Understanding and informing design issues of a prosthetic arm for below elbow amputation by way of "taxonomy" [literature review, reality check]

The academic and industrial attempts to approach prosthetic arms so far have been met with less success than the providers must have hoped for. Far less, in fact so little that we wonder what is going on.

Possibly, design issues are the key to this as however vaguely put, some analytic approach needs to inform better design - but how to really inform better design from issues based on analysis? What is a suitable analysis? If we cannot see any better designs anywhere in practice, real life, then what is the analysis worth? Can we analyze analyses to get a better understanding of what might be going on there?

We might best start with what we know to be true.


Rejection of prosthetic arm option as a clue that something may be wrong, to inform better design

To make better, to improve, means, to take a device or product to a place where it was not before. To understand a configuration, constellation or build in a way that going to a place where it was not before, extraordinary measures are required. It may not be sufficient to lean back and just watch it for a little.

To make a device perform where it has not performed before, one must search to understand and characterize where it has not performed before, by those that know what is relevant there. It is like going to a place to search what is not there in a domain where normative expectations are clearly expressed. That, conceptually, is different from going to a place where you want to know what is there. You definitely need to also know where you want to go.

That means to investigate failure or insufficiency for a particular subgroup of people with a particular mindset, in order to succeed.

That can be uncomfortable and unpleasant, as it is a strenuous issue to connect that type of dots, but it is the nature of engineering that one has to face insufficient and failing function where it is not apparent engineering-wise but orthopedically and functionally, and that one has to expose oneself to it, to make a device better. If anything, one has to introduce relevant aspects of use one wants to see but that were not there before to cover a type of function that is not there any more but in an invisible way. Obviously: because if the problems were all visibile - such as you look at an amputee and you think that you actually understand the problem after the visual of the stump and another visual after putting a prosthetic hand on the stump - then crappy prostheses of the average stock catalog order type would make 100% of the potential clients happy. It is safe to assume that of the 10% of the arm amputees that do wear prostheses, only a fraction is really happy with the device. So you need to get a whole lot of understanding onto the stage before being able to inform better design.

This seems overly obvious.

But here, we will look at academic studies where we wonder what relation to the aforementioned they have. Where the claim they go, i.e., to inform better design, is not followed by going to that step of actually looking into what is missing. So we cannot wonder that there are no better informed designs - a painfully logic consequence.

And while very subtle as a problem, this allows us to re-consider the question, what categories, what dimensions, do or do not lend themselves to actually inform a better design.

Rejection as such is a good point to start this reflection.

Studies to inform about rejection rates: probably not reliable; - forget Google (Scholar), ask me!

An often mentioned aspect is that arm amputees by and large, mostly, reject prosthetic arms. That is important when asking how to design better arms: after all, prosthetic arms are rejected for great reasons. 80 or 90 percent of arm amputees are not all wrong, and assuming that they are all wrong is therefore arrogant.

Studies cite rejection and non-adoption rates as high as 50 or 80 percent or thereabouts.

Of course, when actually confronted with such an aspect of reality, developers or researchers may say that to state such figures "is unfair".

What is unfair in this context?

It is also unfair to say that I need to breathe for air, or, eat to survive. There are some facts that are facts of life. And they are just not fair in a sense that they always empower those that "are" in power.

A realistic rejection rate is a first relevant quantity to inform better design: that analysis alone is indicative of a lot, and it tells you where you want to look deeper, too, and also, where not.

Observation based information about rejection rates

My guess is if you even believe these academically published figures to generalize to a full reality, you should not do research at all.

Then, this all ends here.

Thanks for visiting, bye.

And so by reading on, you decided to reject published rejection rates, be a real researcher and stay with us.

Because in reality, prosthetic arm rejection/ non-adoption rates are a lot higher than the highest published figure.

  • The collective of amputees (below elbow, as example) is an unknown to begin with.
  • But as prosthetic arms are, no arm amputee can effectively hide. If you are in research or development, write this down in your notebook. And if you are an amputee yourself, you definitely know. You just know. They just happen to make the arms so we stick out. It is what they do, for whatever reason that we do not know yet, and it is maybe also was characterizes them. You may have never looked at that way, which is OK. But they make us stick out no matter what.
  • Real life estimates of prosthetic arm rejection rates must be systematically higher than any cited study, as virtually all study authors structurally will be out of touch with most arm amputees, including the ones that never had or that at some point rejected their prosthetic arms.
  • You must acknowledge that even by designing prosthetic device to fail by design, which is another interesting aspect whose reason we do not know yet, you won't reach all of the actual users of your product and that have them tell you off about what you did. Even when facing total dissatisfaction with an expensive consumer product, a large percentage of consumers is known to not do much about it. Logically, structurally, you will not even then get a realistic figure by sitting in a prosthetic center and asking your actual clients.
  • My personal experience of encountering the random arm amputee without or with prosthesis, around here (Switzerland: where national disability insurance would not make it extra hard for anyone to get a prosthesis) points to a rejection rate of prosthetic arms of at least 85%. As my personal observation of this figure covers random street encounters in a society where prosthetic arms are paid for by national disability insurance, basically to anyone that wants one, this has to be regarded as more than just anecdotal evidence. This is probably one of the most precise personal observation based figure. "At least 85%". Safe to go with around 90% though.
  • Moreover, one encounter that I had was with a man who had the same handicap (right below elbow amputee) and that asked me why I would even bother putting up with all that prosthetic arm drama. It was as if there is another reality here, that of "among good unilateral below elbow amputees, friends let friends not suffer prosthetic arms". There is an extra dimension there, that is not just technical. There are dark sides to the dark sides.  However, if you feel like wanting to discuss real rejection rates, then these dark sides become illuminated. It is still not too late to run, though.

So the unknown and the not responsive amputees, including the ones that answer that they do not want to participate in your study, represent the figures, and carry the content, that you really should be interested about if you were to estimate a true rejection rate and if you were to hunt for true reasons. Also, if you want to build arms for every amputee, that is what you want to consider.

Now that a realistic percentage of 80-90% of arm amputees reject prosthetic arms, is that bad? No.

To inform better design of prosthetic arms, let us agree that not asking all those that are not there anyway - and I do not mean "not there" in an existential sense, but as in not there when studies and inquiries are made, not there to apply for a prosthetic arm, not there when needed for the utilitarian purposes of researchers - is a first good step.

Because really, no one also knows, exactly why research and development claim to "want" "better" prosthetic arms based on the general wish to attract the many that do not even wear one because of really good reasons.

That group, in its entirety, may not so much be an interest group but a disinterest group. So either you really want to understand why they are not interested, or you want to build better arms for those that are interested. That is not the same.

And why does research always claim to want to build arms better because acceptance rates are so low? After all, among hard working blue collar workers with unilateral below elbow amputation, acceptance rate of body powered hooks is great. If one focuses on body-powered technology and on adult below elbow amputees that are in the work force, one study [Stürup J, Thyregod H, Jensen J, Retpen J, Boberg G, Rasmussen E, Jensen S. Traumatic amputation of the upper limb: the use of body-powered prostheses and employment consequences. Prosthetics Orthot Int. 1988; 12(1):50–2] reported 10/10 of below elbow dominant arm amputees and 17/19 of all below elbow amputees having become users during a study period of 7 years. In another study, body-powered arms supported a majority of workers also delivering heavy variable work in excess of 8 hours per day [Kejlaa G. Consumer concerns and the functional value of prostheses to upper limb amputees. Prosthetics Orthot Int. 1993; 17(3):157–63], while work load as well as popularity was considerably lower for myoelectric or passive arms. This has not changed since the invention of myoelectric prostheses [Serlin D. The Other Arms Race In: Davis LJ, editor. The Disability Studies Reader. New York: Routledge;2006. p. 49–65].

Research trying to improve or address current design deficiencies of prosthetic arms based on general rejection rates and assumptions into these typically also ignore the total mess that arm amputation really is (it's not just a stump, that would not at all characterize it anywhere near close). In particular, there seems to be no real extra understanding coming from: all "self experiments" where anatomically intact researchers wear extra rubber arms and such.

But if you really want to address the concerns of people that prefer to stay out of the mess of prosthetic arms, getting one built, you would have to address rather basic questions such as these in order to inform better design in a more comprehensive way:

  • Why tend prosthetic technicians tend to treat amputees strangely, at least to a degree where it is a repeated subject in support groups? What is it with them? Why do doctors treat an amputee looking down on them? As amputee, one may be at risk of making relatively disturbing experiences when exposed to the jobs that surround prostheses. Not everyone in that domain is similar, of course, there exist great, polite, correct and proficient prosthetic technicians, but what I heard from colleagues and adding my own experiences, I deeply understand those that wish to avoid this type of human interaction, as ultimately, a very close technician-client relationship will be going to be for an extended period of time once one decides to go that route. Human interaction is essential for getting a prosthetic arm made, improved, serviced or repaired. Inform a better social design there and you have changed the game alogether.
  • Why do customer "service" representatives of Otto Bock, Ossur and Touch Bionics behave questionable, in a way that is unthinkable in other industries? Is it necessary that I deal with these individuals, or are there other ways to go about things? What happens when I avoid that big pile of real problems by just not wearing a prosthesis, or, by delaying any contact by not wearing the prosthesis very often (to make it last a lot longer and reduce repair frequency)?
  • Why is the function only ever so marginally better? Bottom line is that so many people manage pretty well without prosthesis. Why is it necessary to wear one? Why do you think that wearin a prosthetic arm makes a difference for everyone that could wear one, and what would that difference including all disappointments and including all set-backs be for the individual user given their concise environment and situation that often times is totally atuned to the amputee? Can a dialog between researcher and amputee explain that in all detail so we might find aspects to take out for better informed designs?
  • Why is a stump, hosed arm, different arm, so bad that one needs to hide it using a prosthesis? Why cannot you just look away? What was wrong with minding your own business? What image, stereotype, of a human must you have to think that a unilateral below elbow amputee (check blog name now?) necessarily is in need of what you offer as human interaction and as prosthetic arm? Can you inform better prosthetic arm design based on the sociology of looking a bit different? Can you accept that some people believe that one can be a perfectly normal human being without added prosthetic appendage, as a higher goal, and inform design that way?
  • Why does an arm amputee necessarily need to wear anything that is uncomfortable to the point of repeated blisters and rashes? Can you inform design that it better addresses the comfort aspect?
  • Why does the prosthesis literally break all the time, and why is it so expensive? Do they really want to see the amputees coming for repairs every two weeks? Are they that desperate to see us again? Why? What is happening there, really, in reality? Or, alternatively, can you address design and inform it so it becomes more durable? Would you want to? Why not?
  • Who cares? Truly a very relevant question: normal people do not care about amputees. They just don't.
  • Why do you, and why do you so violently and adamantly oppose a hard pragmatic path to prostheses that effectively reduce burden and add degrees of freedom rather than shifting them? Why do you propose complex, unreliable, expensive and painful to wear prostheses and what does proposing these say about you? Is making things extra difficult for amputees what you are really after?

Those amputees that do not express interest in entering that world of prosthetic arm wearing do not primarily want "better arms" in that many questions at that level are also sociological. But, not entirely. Was there a totally trouble free comfortable totally functional fabulous looking arm that is entirely affordable and that lasts long, if ever, I am sure, many more would change over to those that indeed are interested in wearing a prosthesis.

Getting and wearing a prosthetic arm is a distinct way of life

So you surely agree that focusing on those arm amputees that at least have a minimal intrinsic motivation rather than asking pesky questions of the above sort does numerically restrict your audience for very good reasons, but it also bumps up mutual happiness.

As those few amputees that do wear prosthetic arms, of course, we really want "better" arms - but then, what we want, never seemed to have any direct connection to what general research or development produced in terms of ideas, concepts or prototypes. So to focus not on the ones that have to general interest and using vague terms of the wrong (dis)interest group but on the ones that are the (positive) interest group is a very important first step in narrowing the choice of issues to be examined when informing better design based on observation.

Content wise, that is not hard.

Myoelectric arm users want better control reliability, lighter arms, better grips, better suspension and longer function. Body powered arm users hate cable shredding, shoulder harness pressure on the brachial plexus and wriggly wrists as well as weak grips and absent VO/VC/grip hold switches, by and large. But for some reasons, also looking at the latest research, R&D is pussy-footing around these issues, leaving the hot potatoes untouched, for decades now, and quite likely also for the coming decades.

The issue is that we, the ones that wear prosthetic arms, and the ones that go out and beat the hell out of the equipment, that never is enough, are only a tiny fraction of the population, and no real functioning lobby. Research always pretends to want to address "the glorious many that reject prosthetic arms". Wrong focus group, man, I tell you.

Wearing a prosthesis, far away from self-proclaimed professional amateurs in industrial and academic R&D, this is a distinct way of life. It contains pain, sweat, rashes, itch, component failure, bearing the reduced function of failing parts, the ugliness of dirty liners or gloves, and doing away with the delays in getting appointments for fitting, insurance, repairs, and so on and so forth. It is a tedious and also boring path that one goes. It requires both tolerance and intolerance, one needs to react and be able to hold back reaction.

To not wear the prosthesis even as prosthesis aficionado therefore is clearly an experience one makes as part of that. No one understand that reality though. If anything, you may eventually, somehow, be able to actually understand everyday life, grip and load space, as it is accomplished by the successful and happy unilateral below elbow amputee without prosthesis. As that is the main competition with your perceived number one problem: trying to build a prosthesis that somehow succeeds over that type of reality. Even as dedicated prosthetic arm user, you will invariably be super competent in both these areas of experience.

And so we, the arm users, have a distinct way of life that no one looks at. To inform better design, you may want to consider that life style and not the life style of those that were not interested in prostheses to begin with.

Not wearing a prosthetic arm is also a distinct way of life

But they never take a successful user and learn from that. So, you cannot inform any better design by examining an arm amputee NOT wearing a prosthetic arm, and so that idea is dead. Even though it would be entirely relevant from a functional view to see just where things hang.

Not wearing a prosthetic arm, generally, is just as cool as not wearing a large 900 gram camera on my belt, it is just as cool as going to the pool with just a towel over the shoulder rather than with an 18 kg army grade "equipment" bag.  Not wearing an item that is meant to make me socially conform is also both socially cool, and assertive.

So the good old Mambo slogan ("Living comfortably without visible means of support", see image below) is certainly as good today as it was in 1998 when I bought the poster in Melbourne, Australia.

And that brings us to a rarely voiced aspect: arm amputees are exactly like other people, with regard to so many things. This just has to be the most overlooked fact ever, particularly in the imagination of many researchers and developers.

Functionally, my arm stump is not THAT bad, surely for most activities of daily living or in short ADL. Particularly when compared across a free bimanual load distribution across light work and average ADL.

Over 85% of arm amputees are not just wrong! Most arm amputees are people like everyone else - just with parts of one or two arms missing. It is widely assumed that our brains do not work and that we need "a hand back" or "an arm back" to "get back" to life.

But if you, yourself, would not want to carry an extra 7 kg of prosthetic motors, or if you would not want a brain implant, or  if you would not want a bolt causing a chronic wound to permanently ooze pus or other such fluid - then maybe, why do you think I would want that? Can you lay out in all verbose detail why I would really want that, given you not wanting one, so we can all understand?

While a prosthetic arm still may be helpful for heavy or repetitive bi-manual tasks (but even there not a given), it may not be too helpful for all other.

(C) Copyright Mambo

There are definite ways to do stuff with just 1 1/2 arms. With that, we obviously share "tricks" or "techniques" about how to perform manipulations that seem to require two hands, but without prosthesis. Across the years, these things that definitely totally and entirely require two actual hands are not too many. Almost all ADL (activities of daily living) can be performed very well without prosthetic arm on (at which point I remind you, again, of the title of this blog and what exact type of handicap it is I write about).

So there exists this whole sliding grip or bi-manual aspect taxonomy that is alternate, that has ADL and life go by without prosthesis on.

The people you may wish to examine there are the ones who allowed their minds to wrap around that. You want the pros. You want those that never sign up for studies, that are nowhere near clinics or prosthetists. You want the impossible, I can tell from here. At least, you would want that, were you to truly get at least aspects of why people reject or not even consider prostheses. You want to understand just how comfortable and convenient not wearing a prosthetic arm is - how much faster the skin of the stump heals with no daily humid cover, how life is with one or two if not three whole free happy extra weeks per year (that otherwise is spent at the prosthetic repair garage), how life without permanent sweat outbreaks is (did we mention that prosthetic arms make one feel too hot most of the time). And what with the one to three thousand extra bucks of dollars per year, if not thousands to the amount of a few cars or such, what about these? You think I have no idea what to do with these, right? Dream on.

"How to tie shoes" seems to be a really popular subject, and in fact the one that may about be the only real go-getter for prosthetic hand coolness. Operating a coffee machine - as in: pushing a button - or opening a kitchen cupboard, holding a book or newspaper, holding a beer, tying shoe laces with one hand, tying a tie, typing on keyboards, using a mouse, carrying a box, carrying a tray, have been shown to be activities that one can try to complete with a prosthetic hook or hand, particularly with a really good setup and sufficient training. But they all can definitely be, and will be, mastered very well, completely and perfectly well, without prosthesis.

The grip mechanics part with relation to not wearing a prosthesis, however, seems not well understood from viewpoint of those that continuously blare out that prostheses are not sold in sufficient amounts or whatever.

Misconceptions about the prosthetic arm user

Are we happy about heavier prosthetic arms?

There exists this deep misconception and an unspoken assumption among "engineers" and "developers": that arm amputees that wear prosthetic arms are cool with some extra 2-7 kg of stuff, with shoe inserts for foot control, with stickers all over their backs, electrode implants in their brains, repetitive arm surgery for nerve electrode implants, stump bone bolts to stick out through oozing chronic stoma type wounds, and so on.

Without much verbalized or expressed hesitation or doubt therefore, increasing the amount of materials on the prosthesis, and ploughing deeper and deeper into our bodies, has been a typically used (but functionally unsuccessful) way to advance "research" and "development" in prosthetic arm design.

Unless you try out yourself, how a prosthetic arm feels, realistically, you will most likely never understand a word I am saying.  It has to be light and tight.

Are we honest?

Another very deep misconception is that we, as arm amputees, and in that specific role, are always honest.

We are surrounded by prosthetic technicians and industries, on whom we totally depend for parts or support, and that typically act as princes or princesses, that on any occasion may and will take their frustration out on the amputee client - so if ever one wants these gods of control to be merciful and benevolent, one must court them, bow to them and applaud them for anything and everything, and if just to maintain their hopefully merciful and benevolent attitude. This will entail any amount of dishonesty. I have been treated by the customer "service" of several well known large prosthetic companies with profuse amounts of passive-aggressive snotty behaviour, or by total blockage, and not because they are nice or even just competent.

Amputees risk to be treated by any given technician or doctor as entirely, comprehensively, and also mentally incompetent subhumans, so having the gods of control in a better mood often helps to reduce the negative impact of that default attitude.

Also, "helping amputees" that "are inspiring" usually is not that. Not a bit. Sure, we can cover that up, but only for so much. Usually, the "inspiration" spiel works along the lines of interest in staring. People that want to stare, gloat, look, they will call an amputee "inspiring". Conversely, if I want to impress that type of people, the sort that is into staring, I will know what pose is required so they think it is appropriate to call it "inspiring".

Once too much non-sensical engineering is tinkered with and hyped up, and once the big elephant in the room is too big and cannot be applauded away with maximal dishonesty, we will have to address this and ask, why good engineering is not applied. And if ever we will find really interesting reasons for that, who knows.

Arm amputees always buy stuff, right?


Given the extreme prices of "bionic" hands, a few entrepreneurial amputees started to "advertise" or "promote" these devices as "ambassadors" or "key users" or maybe as "sales pre-force". They pose as "inspirational" models, they "inspire". Obviously that euphemism has different real connotations, so the prosthetic industry attracts people with an interest in staring, and clever amputees can exploit that for their benefit, so there is the business for mutual benefit. The amputee poses, "inspires", and gets a "bionic" device for "free" by paying for it with their advertising, basically, and the company gets free advertising with amputees that pose for them and that "inspire" others by, say, holding a beer, holding a newspaper, or tying a shoe - all of which we all also do perfectly well without prosthesis on.

Of course, we all know that these devices never work as advertised (otherwise the advertising would not really be all that necessary), so that advertising as such can be relativeöy hard work.

Anyone that even manages to pose with a "bionic" hand while rock climbing must be applauded, and even if just for the audacity. One iLimb ambassador in Zurich trying to give a workshop once said in the afternoon, after the workshop, that her stump hurt, and indeed when she took off her prosthesis, it was all red - but she had not one bit lifted anything heavy or manipulated anything extreme. So there is also clear suffering offered for the price of free prosthetic devices that are otherwise impossible to pay for. I don't suffer nearly as much wearing my body powered arm, when doing real work, even, but then I don't pose as model for a prosthetic arm or "inspire" others in the same degree.

And  trustworthy rumor has it, that not all "ambassadors" really are that much into these gadget arms, and we even could start trading names - but we won't. They represent, and they advertise for a bit, and then that's it.

This here is not commercial advertising, either. I am far too independent and irreverend to praise an item for ambassador purposes. I have been asked a few times by larger investors whether we can turn my blog site into a commercial advertising thing. And I declined. I am no one's ambassador, and when any piece requires improvement this will be very apparent in my writings.

Summary of focus considerations: focus on current prosthetic arm users!

It would be enough  to focus on trouble shooting current prosthetic arm users. Going down that very long path together with a prosthetic provider, to steer towards a successful prosthetic fitting, and then to use that arm purposefully, that certainly takes up so much effort and time that we have to call it a distinct way of life. Honor that, will you.

Arm amputees as research rabbits

Wearing a prosthetic arm is a way of life. It requires a lot of perseverance.

And then, one step further into suffering, there is the life of a prosthetic technology research rabbit.

Most amputee research is totally harmless, from view of a research rabbit. I participated in a few studies myself. With one exception, they were not painful, and often, talking and writing was the most important part. I would not consider open surgery, however, as example of what would not be acceptable for me. Otherwise, being a research rabbit, I found that most results were practically useless, without any impact whatsoever on my own prosthetic use or equipment.

As a researcher, having research rabbits cooperate in non-painful or information type research may not be enough. As that, you would need a rabbit for more extensive tests, maybe not to totally slaughter right away, sure.

But to go a lot further.

And you will invariably find the single amputee that does all that you want - that gladly suffers total surgery, that complies with 8 hours therapy a day for a few years, that wears prosthetic arms with some extra 2-7 kg of stuff, with shoe inserts for foot control, with stickers all over their backs, electrode implants in their brains, repetitive arm surgery for nerve electrode implants, stump bone bolts to stick out through oozing chronic stoma type wounds, and so on, as stated above, all that.

I will not say that no amputee that is right in their head will agree also to extensive surgery and so on.

But as a default attitude, amputation may go along with serious depression, and depression does the weirdest things to people. So it is not a surprise to see amputees offer themselves to be operated on, experimented on, "used" for research. Some amputees will say that they want to "be useful at least for something" - as if a utilitarian attitude then beats depression and disappointment.

As a normal reaction, the majority of arm amputees will go "whoa" when the offer is, to, for example, accept an experimental brain implant. Most of us, particularly after the first disappointment fades, are well adjusted and careful about optional experimental risky try-outs.

And we may be somewhat desperate, with little or no way out: myself, I obtained ethical approval to perform a self-experiment at work using different prosthetic setups, some of which caused abrasions, friction rashes, blisters and other negative side effects.

There is no sensible other way out - I will want to work and I will have to work no matter what, so in a way, that little research rabbit is cornered. With that there is no other choice - whether the results are published or not. Show me an amputee that is entirely free in their decision, and not under immense social and economical pressure, and I show you someone that just won the lottery.

And so if you want to perform research on an amputee that otherwise feels useless and wants to contribute to humankind by having you slice him up and tinker around with him, maybe consider this to be a possibly bad idea ethically. One can ask a psychiatrist for an opinion, possibly that helps clear the situation. But overall, these can be very difficult and critical situations from an ethical view.

So if you want to inform better design, maybe restrict yourself to those research rabbits that are free to consent, that have undergone a short psychiatric assessment, and where participation in research is not their only chance to be important, do something meaningful, or their only chance to get a prosthetic fitting.

How to simulate the feeling of wearing a prosthetic arm

To walk in another persons' shoes is a great eye opener, but only if done right. Developers, engineers and other people tinkering their way towards my reality or similar realities risk do not understand the domain from bottom up.

And trust me, you want to really see that angle from frog eye view, so to say, from bottom up. As stated above, it is far more of a clusterfuck, not "just a stump". And so any simulation should also aim to screw things up far more comprehensively than to just immobilize your hand.

Or as I would say, sweat, itch and skin rash first, control and grip failure later. Then, absolutely insane cost, personal efforts and wasting a real lot of your time on this. Public humiliation last. You must experience the hell aspects to estimate what all goes into "wearing prosthetic arms as distinct way of life". And what all goes into being liberated from that. By not wearing a prosthesis. So and so many arm amputees really suffer from depression, and there must be a way to actually get you there, too.

If you are not there yet, then, you are not there yet.

First, you would need a neoprene sleeve to cover your arm at least to the elbow or maybe even over it. The neoprene sleeve will have to be tight. You will want to sweat underneath it, and you will sweat also because the thing will heat you up.

You will wear it 14 hours a day, 7 days a week, for at least three months.

I mean, others complain they don't sell prosthetic arms, as if it was like selling carrots or hardwood floors - and seeing as if one costs maybe 80'000 CHF (like, an iLimb), you need to know this. So, off you go, wearing this.

After a day or a week or so, you may get a sweat rash - lots of small red dots, crazy itch, skin oozing. That is normal, super, success in simulation and of course quite irritating and very uncomfortable. Put zinc creme on it if you want, and keep trucking. This simulates suspension, at least a bit. It should let you know that wearing a prosthetic arm often tends to be rather uncomfortable as long as you do not sit still. This will then teach you, maybe, to sit still, who knows. Then you should really see what keeping a heavy weight burdened arm very still does to you in the long run.

Secondly, your hand can only open and close. Nothing fancy. Use elastic bandages to make your hand only grip but not use single fingers. Maybe add shrink wrap to it, add elastic bandages, make it as reduced, grip wise, as an Otto Bock system hand with a fixed tripod grip. Maybe see if you can get a cosmetic glove over your hand that is as thin as the iLimb glove. You need to get a chance to experience situations when the cover tears - then you spend 700 CHF on a new hand cover. Sure I can perforate a  normal iLimb glove with 10 minutes of car wash hose holding - and there is no reason you should be any better off in any simulation.  Of course, you may easily end up spending a few thousand bucks on gloves, but, then, you really need to experience simulated reality. Pay for it. Give that money to charity but make sure you really give it away - otherwise you cannot understand people that steer away of such a lifestyle. Walking in my shoes is not thinking about walking in my shoes. It is "walking". Hence the words.

Thirdly, wear at least a 500 g wrist band. It simulates the extra weight of the prosthesis, and mind you, 500 g is not much. For realistic myoelectric weight, you could easily simulate it using a 750 or 1000 g wrist weight.

Fourthly, restrict your elbow motion somehow in order to effectively simulate a Munster style socket, the most prevalent socket style for myoelectric arms. Maybe use elastic bandage around the elbow. After a week or two of typing with this setup, your shoulders and neck should be painful and tense, and after a month we want you to feel iron hard muscle pain and irresolvable tension. Then you may start to get a feel for what goes on with a prosthetic arm.

Last but not the least, you want to experience the true Otto Bock part imposed (or other company parts' imposed) cable shredding: mount cable, work stuff, have cable rip suddenly whenever it feels like failing on you, then, take things from there.

Then you wear that, at least three months, seven days a week, fourteen hours a day.

Before you get into failing tasks such as carrying or manipulating items, suspension on the skin may be the ultimate fun killer.

To experience some more serious exposure, with that type of setup, go and cut hedges in 35 deg C summer heat with sun for 2 hours. Go install a 550kg total weight wardrobe system in your bed room. Bike up the Stelvio pass on your mountain bike. Do what I did, and tell me how it was for you. Feeling this being a nightmare? Hm, was it your idea to sell more of these arms to people? Are people that do that to amputees now still real humans to you if I may ask?

Ultimately, the shoulders and neck affected by asymmetry and prosthetic overweight will also wear you down. You should suffer that for a while, as this has the potential to greatly expand your understanding of what people want to wear or not.

Once this has kicked in, never believe it could not be a lot worse.

Keep going for a full three months as only then will you have a good chance to get a good experience of what it may feel like. If that does not do it for you, try a whole year. And trust me, one year is nothing compared to, say, five or even ten, to sharpen your senses and steer your intent. It should really clear your mind from all doubts.

Then, consider the loops and hoops that an arm amputee has to go through to get these things financed, repaired, maintained, replaced, to get training and to manage back-up plans. The best way to test this, 1:1, is to spend one full afternoon of 3 hours sitting in a room without windows and without any web connectivity. Just take every Thursday or Monday afternoon; you will excuse yourself from meetings or deadlines as that is just what wearing built-to-fail prostheses means.

The times when you can not wear a prosthetic arm, or decide to give that poor thing some rest, will then contain a hint of true freedom. The grip aspect of working through life as right below elbow amputee without prosthetic on are not just that bad - we will get to that in just a few moments, below. But the sudden lack of the accumulated weight of issues is not a neutral, unemotional thing - no. It is so liberating that you wonder how to suppress that joy. And to inform better design that gets this angle, a totally different taxonomy of crazy stuff is needed than a taxonomy of whether three, four or five fingers are used to hold a cup.  And as things are, there are no better designs, as too many people remain totally uninformed with regard to that.

So, it will take a lot of great aspects of the prosthesis, to make a unilateral below elbow amputee really want to wear or request a prosthetic arm. To inform better design, a first must-have baseline from my view is a minimum of three months escalated suffering as described above, I am very sorry to say. But, if I survived and others did, whom you may prefer to look down on, we can only reach eye level again once your body takes that in to a degree where you know what I am saying before I say it about this subject. And I am hopeful that we can get you there. It is just a bit of hauling ass until better informed design can be given. But that is alright, we waited so long so far, we can wait a little longer.

Categorization as way out of a mess?

Jesus is coming - look busy!

It is only possible to fail your clients, over decades, if manufacturer or researcher and user realities are sufficiently disjunct. After the disjunct quality of materials and user expectations is established for a sufficient amount of time, proving the cemented nature of the power relations, it clearly is not a primary focus to just build "better arms" but to understand what the underlying sociological process is that goes awry so reliably.

This must be rather disturbing as a fact for those that "want to help amputees" but it must be great news as well - exposure to real life, real work and real amputees where many of us are sweaty, needy, overworked and suffering from constant pain may not be all the glorious fun you imagined to have as successful academic or industrial researcher. So in a symbolic way, you want to perform clean and successful work, i.e.., save the world by not touching it. And that is almost impossible, unless you find sociological wormholes.

But that has not been found yet. And the feeling to be almost totally out of control may be disturbing for those that represent research and development with the specific premise of trying to make prosthetic arms better for better acceptance.

On the other hand, the longer you can draw out actual resolutions of actual problems that prosthetic arms pose, the longer you may be able to keep trucking, pretending to solve problems and such.

Imagine if there was a single affordable end-all be-all prosthetic arm that would cover all aspects, like, tomorrow. Nothing would be left to do for you, and we could not have that either.

What better than to start by "cleaning up", "sorting", "categorizing", bringing order into what may appear to be a somewhat uninformed total chaos and mess?

Any attempt to "sort" the "mess" that arm amputees seem to reside in, by using prosthetic arms in ways that (a) are different from how human hands are used and (b) that are different from how one would expect it, seems very promising. So let any obscure categorization begin!

Such an attempt has the theoretical capacity to heal the disturbed if not wounded researcher that suffers from the apparent mess going on on Planet AA, wandering through deserts void of inspiration, howevermuch isolated on any outpost on Planet RD.

Finally, let someone sort the mess! Sounds very straightforward.

Disjunct worlds

As we will see, disjunct worlds as main aspect for research and development versus real life continue throughout more refined instances of the same problem: what happens there, on research, development and non-disabled planet (let us call this Planet RD, as for research and development) may just not have a lot of bearing on what we may call Planet AA (arm amputee).

And the amount of perseverance that we see there, combined with the fact that we are not really moving a bit conceptually (are we?) tells us that this may very well be with cold intent. Just a thought.

The intricacies of truth saying have it, that truth saying on either Planet is asymmetric in relation to the other planet.

So any true inhabitant of Planet RD embodies the perceived truth that Planet AA people are infectious, mentally incapacitated, sub-human if human at all (due to lack of hand, arm part, due to communication relevant disfigurement) and definitely not credible, able to speak for themselves, and such.

This makes posts such as this a technical impossibility. And with that, you, as reader, become a logical impossibility - I know; Rene Descartes once ordered a beer at a bar and when the bartender returned asking 'was it you that ordered the wine' and Descartes replied 'I think not' - when *shwoop* Descartes disappeared (how is that funny: Descartes is quoted for "I think therefore I am").

There have been papers that "warn" researchers from "reality" - and I admit, this may be a total affront from view of an amputee on Planet AA, but it is only consistent, and true, to a Planet RD centered world view. The rules on Planet RD must be along these lines:

  1. You must cite some "reason" for your research. The usual "reasons" are "amputees reject prosthetic arms", "prosthetic arms cannot feel and grips are strange so we need to implement feedback and better grips", and, "prosthetic arms are by for not as sophisticated as modern technology allows so we can now go ahead and stuff all technology into the prosthetic arm of the future". Cite whatever you like, just write there that this research that you are doing is really needed. As it will bring the lost >90% back, or was it >85%.
  2. You must then provide massive electronics. Modern prosthetic arm research must be electronic. Just after the first rule to mis-cite amputee studies to allow any crap on your prosthetic arms, you must totally submit to the electronics credo. "Anyone that does not use  and push electronic myoelectric arms is obviously a moron". You could also use some flashy statistics - there, go by the "mesmerized bunny" mechanism to earn your credits.
  3. After that, you can basically do whatever you please. The sky is your limit.
  4. Never actually discuss your concepts with arm amputees. Tell them your concepts, talk down to them, treat them as subhumans that they are regarded as on Planet RD (being as disfigured as they are), but do not ask them for opinions - instead, talk top down and be the authoritative voice you always wanted to be. Avoid face offs and avoid discussions. If it must be a "discussion", pick an arm amputee that plays by your rules, preferably one that has a contract with a company to sell Touch Bionics iLimb or an Otto Bock Bebionic hand in role of an "ambassador" (i.e., sales rep). Or one that has very low self esteem and that believes being a research object or promoting "bionic" arms helps "their value".

Conversely, there are strict rules on Planet AA as well. They are just as wicked.

  1. Live life with a low footprint with regard to prosthetic arms. Either go without, or, get a conventional prosthetic hook - these are actually helpful for bench work, garden work, car work or other stuff and they hold up a tad bit better than myoelectric crap. The less you use the prosthetic arm however, the less chance it has to break. And the less exposure you get to the whole apparatus of insurance, technicians, repairs, spending so much time and so on.
  2. When anyone, any one, including prosthetic technicians, doctors, Planet RD people, family or others ask why you do what you do (see rule 1) tell them some avoiding or lowly rubbish such as prosthetic arms cannot feel, grips are weird, etc (and feed them the nonsense that goes into rule 1 on Planet RD). Out of sight is out of mind. You do not want Planet RD to further the development of less objectionable but even more uncomfortable stuff compared to what they build now already. You would want a light great solution but for that they'd have to listen and to prevent that they have hard rules in place. As that is not hard to recognize, any evasion that saves face and allows everyone to tucker on will be cool.
  3. Keep cost, efforts and material down. Learn tricks at that level. Wait, be patient, stick it out. Then let things happen by themselves. This works so much better than anything else. Avoid any head on collision with insurances, manufacturers, technicians or product ambassadors (amputees that act as advertising representatives to get free or cheaper parts). Mum is the word.
  4. If you actually need a true solution for your prosthetic arm, go outside existing industries (just as long as they reside on Planet RD).
  5. Go and use the materials to the fullest.
  6. The attention that you get because of the handicap is a real trap. It is not personal attention but a type of raw neurological attention that others cannot handle well (on more than one account, actually). It may not be beneficial to take advantage of this type of attention.

With a bit of relaxed cynicism, you can easily see how both of these rule sets, if applied to both Planet RD and AA for the past twenty years, describe the current reality fairly well.

As this website implies, there may be a third way and there, we may have to shine some light into the nonsense that is going on on both planets.

Categorization in general

It is just so unfortunate that categorization as such already is a tricky subject even on a single conjoined world, leave alone considering separate disjunct worlds.

Generally, any space, domain or area of application such as "grasps or grips in everyday life" can be categorized or analyzed.

It seems to be tempting to apply some type of "obvious" categorization scheme for human grasps, to just use what others tried before.

As grips historically were classified into some object-gripper relation, the problem may be to find out in what relation the actual categories and historical categories of grip classifications are, and in what way any of these inform a better design. Possibly, these categories possess no informative value.

As any categorization or analysis usually does not cover unlimited dimensions, we need to start with what is commonly referred to as purpose or goal, as target or vision.

This is a very general aspect of all categorizations or makings of groups.

It is a fallacy to believe that categories are obvious or self-explanatory.

Even IQ tests may build on such categorial or sequential order assumptions that are, if one were to more harshly look at the question, ill founded. From [1] cited from [2], a train diagram serves to illustrate this quite well.

(C) Copyright

The train classification experiment posed three questions:

1. arrange trains into any number of groups, 2. into  two groups, 3. into any number of conceptually similar plus an "other" group (details see cited references).

With 31 participants, 93 classifications resulted, 40 of which were unique.

In other words, there was a wide range of singleton solutions.

As a critical aspect, only disjoint rather than intersecting classifications were made.

People, left by themselves, confronted with the mess of unsorted trains, asked these questions, resorted to non-intersecting classifications.

If nothing else, this illustrates that people that are generally uninformed about the wider issues of categorization tend to not properly justify their categories (but provide willful, unfounded, unexplained, not derived but as-is claims).

They tend to assume that categories need to be disjunct, as "square vs. circle", "large vs. small" are typical disjunct categories.

This is relevant inasmuch as the example of trains (image, above) does contain many overlapping aspects.

The actual issue of categorization from a more generic and pragmatic perspective is that

  • if it lacks dedicated, focused and targeted purpose, it remains pointless;
  • if it lacks adaptation to the domain it must work in (and that may well contain overlaps), then it also remains pointless;
  • if it lacks a normative framework, it may be pointless;
  • it may require a type of class definition that relates to a target, to a goal, to an outside reference, rather than being sufficient when based on presented items only;
  • in order to provide statistical or analytical approaches that reside inside meaningful categories of use, of failure, of goal and of improvement, you must already have an inside understanding of such dimensions.

So the real cognitive problem lies in creating a meaningful hermeneutic circle that breaks across Planet RD and Planet AA, given strict, hard, non-compatible and non-intersecting rule sets that are imposed on both planets, and that relates to the real world in many aspects.

We must not forget that the actual goal is to build better prosthetic grippers. That means that from that perception alone, current grippers are not entirely satisfactory.

  • See what people with two hands do that arm amputees do not or poorly. Build a gripper that solves such a problem.
  • See what no one does, either with two hands or with an arm amputation - but that would be cool to do, or necessary, helpful, or possible. Build a device to do such a thing. Such as shooting fireworks.
  • See what current prostheses do but poorly. Usually there, functional design, loads and forces and materials, technical design and construction as well as manufacturing go hand in hand (bwahaha).
  • See what arm amputees do well - give or take - without prosthesis compared to with prosthesis. Take it from there with better design.

The list is probably not complete, there must be other aspects.  I can feel it.

But if anything, the list clearly says one thing very loud and clear: before you are an absolute Jedi master at design and construction, you probably will have a  hard time advancing prosthetic grippers based on a few simple unreflected categories.

Attempts to categorize prosthetic grips: Adam Spiers, Linda Resnik and Aaron M. Dollar (2007) Analyzing At-Home Prosthesis Use in Unilateral Upper-Limb Amputees to Inform Treatment & Device Design

The declared purpose of this paper [3] is proposed to be "for better device design, assignment and training".

What the study does is to use three different prosthetic arms, and it provides a categorization of the grips that the users employed.

This paper employs a focus on a type of categorization that is not too irrelevant from both an actual user perspective, and more so, that wishes to be relevant from perspective of improving current designs.

To really inform better design, however, comparing current designs with applications that the users should provide but do not provide for design reasons seems to be a more rational approach.

Image from [3] :

(C) Copyright

The categories seem make an impression of trying to be disjunct, so, no overlap to be assumed there.

  • "Bilateral action" appears to be an ill defined category that is side by side with "bilateral carry" and "bilateral transfer"; how do "bilateral carry" and "bilateral transfer between hands" not represent "actions"? What about holding an item, or taking a photograph using also subtle manipulations?
  • Why is there not one category called "hold and defy gravity" - it then would also comprise "pinch", "power" and a few others?
  • Where on this earth is there no gravity to be defied?
  • Why is "other prehensile" categorized underneath or inside "pinch" when really it is disjunct from "pinch" with an image showing something like (but not totally a) "lateral" grip?
  • Why is "pinch between fingers" not outside pinch and termed "jammed between fingers"? Isn't that more a lateral between-finger grip than an actual pinch, technically speaking? Or is that not a technical paper after all? As depending on the hand model, the interdigital space varies, fingers are flat elongated and cylindric on their lateral sides, and no pinching grasp or mechanism is involved for object stabilization there? How are "push" and "stabilize / support body" geometrically different?
  • Where is typing? Typing, as distinct push action?

The authors of this 2017 study recommend nothing less than "designers may wish to add compliant finger-pad-like surfaces and tactile features to the outside of TD fingers, to facilitate object pushing and clamping".

  • The V2P Prehensor came out of first developments in 2008, I was happy to be involved in discussing features then, and without long agonizing video loops or categorization attempts, the feature has been already there now since almost 10 years, undetected by this research, as planet RD and AA are disjunct. I told you this was relevant. They have no idea their new research result is already on the market. That design had already been "informed" by simply breaking with rules 2 and 4 of Planet RD.
  • The actual use problems with pushes are totally elsewhere though. If one is to use this really, day in and day out. Really, proficient push use also contains heavy typing. Ultimately, push area shape, hard or soft quality, size and angle are absolute key to better "inform" design. A simple consideration of actual use of prosthetic arms could have yielded that information.

The authors of that 2017 paper then proceed to state "Furthermore, strengthening lateral stress capability, or adding compliant mechanisms in TD fingers could enable further non-prehensile hanging, clamping and pushing without fear of damaging the TD".

  • The statement is in itself not wrong. But, no shit. Honestly. Did you seriously believe that I will wear a ~1200 USD priced Hosmer work hook with a ball bearing and made from solid steel, and NOT naturally and impulsively use that to rip open cartons, bang in dowels or nails, you really think that there is any fear of damaging a TD by slamming it forward, backward of sideways ("laterally")? What was this so far and up to now: a mannequin sit still competition? "Don't exert yourself, the hook may break"? "Wait, we have to inform design"?
  • Let these designs be informed that we have taken our business elsewhere, altogether, and that we have started to use these also to not just tinker but provide solid mounts for those furniture sets that safely qualify as "Lego from Hell".
  • There does exist this land, where use and abuse is the same. Even though it may be on a different planet than Planet RD.
  • And besides that, not even a regular Otto Bock hand died because of lateral push forces. They died because of the mechanical design of their pull mechanism. If you really want to improve current designs however, then, a different type of approach will be required.

This paper raises further questions along two lines of conceptual understanding:

  • Lack of force quantification: If it ever was for making existing devices better in terms of quality control, reliability testing would be necessary to inform better device design. This also entails high load testing. The results would then have informed the authors that load is absolutely everything. If anything makes a relevant difference with relation to design, it is the load. Up to a low grip force level, mode of control (VC, VO) can likely be neglected, but not in excess of around 6-10 kg of gripper force. Up to a certain low weight, it does not matter whether I balance an item on top of the prosthetic hand or whether I grip it; from a certain weight upward, balancing becomes difficult or impossible. So grip, gripper design and quantification of forces are intrinsically linked.
  • Lack of consideration of failures and non-performances: To provide a better device design beyond current designs, a study must identify areas of failure, non-performance or absent use of current designs in relation to expected or necessary grips or activities.
  • Particularly, it is epistemologically impossible to just look at video footage of an arm amputee performing stuff and on basis of just that identify what all is missing unless a normative expectation is expressed, defined, present or set, that the video is compared against. So, the list of items not performed or failed is missing by study design.

Actual grip focused issues are quite insiduous.

  • From a practical view, I will distinguish "bilateral grip with a firm grip on both sides" as when holding a glass or bottle, from "bilateral grip with a locked grip but sliding / swiveling enclosure" needed for vacuum cleaning, turning the handle bar of an awning, window blinds or operating a heavy hedge cutter or lawn mower. The paper does not give this design-relevant distinction; if it would have done that, it would have explained the reason of existence of a Hosmer model 5 versus 6 (work hook) design difference, and where Trautman and derivative style hooks would be better or worse. So a study that does not explain the existence of proven grip difference needs of successful hand or hook builds cannot possibly go further to inform "better" design.


  • My prosthetic wrist started to come lose early into typing on a keyboard. The usual way for many users to deal with this is to not type that much. Had I started to not type that much, no engineer would ever have been able to categorize my daily grasps and use it to inform better device design. The real problem is that useful wrists are not developed if one does not know the domain of prosthetic arms in usage. So, typing as an entity is not helping to inform better device design unless it is specified as "the type of vigorous, extensive and drawn out button push operation that does not just non-grip/push a single button as such but that strains wrist, socket and stump skin as well unless addressed by better design". Quantity and forces are everything within that distinction - if, at all, better design is what is sought.
  • My cables ripped every 4-10 days with the grip activity I had. A normal reaction of a more relaxed user is to reduce that activity so the cable remains intact for 3-5 months. Had I started to chill and reduce grip use to conform to the bad cable sheath mount design, as a user that sees the "researchers" failing to recognize actual issues as that being what they do, no developer" would ever have been able to identify this problem by segmenting my daily activities given even massive amounts of video footage. Now: I went out of my way, and designed and built a different way to mount cables - now they stay intact, unserviced, under full load, for over 9 months. That extended my range of activities considerably. I can now grip a lot harder - and it is a control cable design issue that allowed it. Only the specific identification and designation of "grips in an amount, and with cable loads, high enough to kill a unilateral cable sheath bend within 4-10 days" as unique entity is capable to inform better device design - and I know because I informed my device designs on exactly these precisely defined entities.
  • My hook can be used to make coffee. My Becker hand can be used to make coffees. I can make coffees without prosthetic arm on. I may go about it slightly differently. But that is not even important. The really important activities to focus device design on are the ones with a heavily repetitive or large weight aspect and the ones where bi-manual failure is not at all an option. An unweighted segmentation of a video of all my grips over a full day does not do this any justice. One real problem here is that asymmetry takes a toll on my back, not from the few coffees that I make, but from repetitive heavy lifting or long drawn out typing. So forces, dimensions and weights of objects in contrast with the success of the  attempt need to be weighed against the importance of the task. There lies the potential to inform better device design with weight on relevance.A particular prosthetic hook may be sufficiently strong to lift a 0.5 kg item but not to lift a 5 kg item, but the necessity to improve a design feature there depends on the importance this task has for the user. This plays out, in everyday life, against all alternatives the amputee has. Device design may be not possible for most, but tricks, work-arounds and techniques without the prosthetic arm or with a deficient, weak or otherwise poorly designed arm may well be an option for many. To design a device that performs above my baseline performance with a stump, without prosthesis on, is where it can make a difference.
  • If you list a range of daily activity aspects where people with two hands thoughtlessly perform extremely well (and amputees do not), then we can, off the back of our hands, list sexual activity (and if you have no idea why two hands are so much more important than one, you need a lot of help far beyond the scope of this), handling complex wallet / cell phone / key situations with almost perfect grip reliability (ever drop a cell phone?), hopping on a moped or bike, and playing instruments - all with ease and elegance. On top, they never look as if missing a hand. So if you ever feel in need to actually inform your device design to perform or look better, come here first. Ask people that know what better device design *is*.

In other words: to inform better device design, the design deficiencies of current designs in context of the actual use are one thing (and may mostly pertain to creative off-label variation use or reliability testing), but the design deficiencies in context of intended, necessary, required, planned, hoped for or enabled new uses weighed by importance, and weighed by orthopedic long term impact, are how really better design is informed.

This study does not address these aspects in any user or engineering relevant way. Therefore, the study remains irrelevant on Planet AA. It is a great and shining example for a lost cause, however.

Attempts to categorize prosthetic grips: Belter, Joseph T., Bo C. Reynolds, and Aaron M. Dollar (2014) Grasp and force based taxonomy of split-hook prosthetic terminal devices.

A similar paper appeared in 2014 [4]. The authors state that they were interested in specific design shortcomings of the split hook.

As a rather heavy user, I can tell you for free that the specific design short comings of a split hook are:

  • inability to have 1 split hook shape to fit all occasions, hence the need for different shapes;
  • inability to have force of VO devices always be correct, hence the need for both switching forces and maybe also switching to a VC option or device;
  • the need for some hard and sliding grip surfaces for some uses (such as hard curve driving and steering wheel manipulation) and soft deformable tight grip surfaces that rapidly wear down for other uses (and a need to switch these, fast, and cheaply);
  • sometimes a too high pressure as in force per area, while loads are obviously high;
  • implicitly, cable sheath wear down of conventionally installed cable housing was the single most predominant issue for both force limitation and force dosage or fine control.
  • Wrist stability is a major factor in very repetitive use of higher impact or pull forces.

How the taxonomy was generated is not explained in actual detail: so, one could get some grip statistics across various categories of uses, and resulting grip situations then can be normalized or reduced to yield relevant different domains or such. No - here, what lead to the taxonomy is not clear, except that the authors performed "thorough" discussions. Yes, this here is also thorough, trust me.

This certainly sets a precedence for the type of scientific approach that is regarded as valid there: discussions, if they are "thorough", are what is totally accepted.

The head camera retrieved video footage was split into what appeared to be non-overlapping taxonomies of grips and force applications. The taxonomy was then applied, and that was that.

Grasp taxonomy

The "grasp" taxonomy contains a number of categories.

Image from [4]:

(C) Copyright

Non-prehensile tip. This category falls into two major applications: iron hard or at least very steadfast highly repetitive industrial type hammering, typing or such, bringing conventional wrists to early failure, and, soft touches. Engineering wise, the heavy tip category is specific and needs to be highlighted - simply because there are designs to be informed, improvements to be designed. Heavy tip pushes as in industrial level keyboard typing usually causes wrist and wrist mount to fall apart, but this requires this particular application of that grip type to be individually termed and tagged. Further down the text, and also since then, these authors have not identified wrist and wrist mount wriggle as key issue in heavy tip push usage, so they never informed better design based in that grip type. That is why amputees like myself end up having to build our own wrist, as example. Failure does have an anatomy, too, sometimes, that can be dissected.

Non-prehensive front. The same applies here: very heavy pushes will require a specific design improvement, which is entirely non-apparent for plastic cups (as shown here). No one will perform a very heavy push with a split hook, as that apparently won't work. So you can probably segment video loops of compliant hook users for a few decades or so, and won't ever understand this problem.

What is needed, as case example, is to push a 70 kg body from one striker or table to the other is a large flat wide surface. One can look at the V2P prehensor and see what the designer understood there. Look at other amputees that do perform heavy pushes and tell me what you found. Were they wearing a TRS Dragon device for pushups or other training? Or not even a prosthetic arm at all?

So, design improvements won't ever be apparent unless force, heavy use, exceeding current model specifications, are considered as very specific usage variations. Or let us put it reversely: if  conventional current large firm stock model devices fail some application, then, any category taxonomy to inform better designs must necessarily map the problematic use aspects as specific, out-standing and recognizable entity. Mild light gentle pushes are not informative as category - not if they are also possible to do with a soft boiled spaghetti. I exaggerate only mildly.

Non-prehensile hook: the image implies that a large and thus possibly heavy suitcase is lifted. Any design that has to be better informed invariably will touch the quality of suspension and of the whole socket, wrist connector and adapter mount. The paper disregards this. This insight is fully dependent on quantitative aspects of the lift activity: lifting a few hundred grams never poses problems. Once a certain weight is exceeded, a grip may not do, and wrapping a strap may be the way to ensure lift force transmission. So the grip category as such is not informing better design for the intended act - lifting by a handle, that is.

Prehensile two-contact grips. When lifting parts off a surface, then the slanted or straight geometry of the grippers, a possibly soft surface design, and other intricacies are very relevant sub-category features. The design for optimal work function has not been systematically studied there. Repetitive work as in repair or construction type work, heavily relies on fast reliable small object grips. It is my anecdotal experience that a somewhat wide soft deformable gripper surface is the best option there. However, the subspace of unmodified, modified and additionally ideal gripper shape and surface design including material specifications has not been properly illuminated so far. Naming the fact that objects are gripped using a two-contact grip is not too revealing though.

A missing hook or gripper application is the lift grip that goes underneath a flat box such as a hand with rectangular finger flexing. This is a very important grip type, possibly related to the non-prehensile hook grip implied to go underneath a bar shaped handle, but without notch, edge or other grip-stabilizing structure. Also there, weight is the key issue design wise. For higher weights, the curved shape of some hooks in conjunction with hard surfaces of the lifted object may become a real problem. Just try Hosmer models 5 versus 6 when carrying a heavy 25 kg computer workstation device. You need massive extra force with the wrong hook shape there. As this difference needs to urgently inform better design, forgive that this has already been done. A careful examination of the V2P Prehensor features will reveal how it is ideal for heavy hard object lifting with exactly this problem in mind.

Prehensile - Here, the proposed distinction is academic. As everyday objects are somewhat random shaped, the hook grip just encounters them accidentally. If anything, a generic approach to approximate an ideal rigid two-claw gripper shape based on any number of given objects would be interesting. Then we could feed everyone's different list of everyday objects into such a generic approach, and a useful hook design with specific adaptation would result.

From a two point to many-point grip, the soft surface aspect is relevant in securing lower point count grips. My two-point stump elbow grip is geometrically simple as it seems - but the soft skin provides form closure for many differently shaped and sized objects, which a metal hook won't just provide.

A sliding encircling hold (but not fix within the grasp) use has not been defined, but is a reality for vacuum cleaning (full sliding of tube), brooms or shovels (swivel motion free), driving a car (swivel grip when turning steering wheel) and cleaning up cables (rolling up a long power cable into loops works best with such a grip).

As we grip also soft objects such as fabric or PVC covers that then need to be pulled, such a taxonomy to inform better design for that also is missing - as the problem, when you pull the PVC cover on which a body of 70 kg lies, and the PVC then tears up with the hook tips, is real. It is best remedied by larger contact area of the hook, which distributes forces across a larger area and thus reduces pressure on the object material.

At the same time the focus for gripping or grasping heavy and large items should not restrict itself to terminal devices.

When I carry really large parts, I usually take the prosthesis off and bear hug the item. A prosthesis for very large items would need to also use the prosthetic socket as added grip surface. So better design would contain a rubber padded socket inside for better bear hug lifts. No one, ever, considers this, ever.

Force taxonomy

Image from [4]:

(C) Copyright

There is considerable overlap with the above taxonomy, particularly with a goal to inform better hook design.

Pushing is not such a great category for any taxonomy that wants to inform better design.

Hard pushing maybe is.

  • One device that was informedly built for very hard pushing is named TRS Dragon. I have one. It is really good.
  • For everyday use, the V2P Prehensor has good push surface features. There are soft ridges that are specific for pushes.
  • The typing performed with any hook device on a suitably robust keyboard also falls under the push category. After some 2 months any epoxy socket mount and wrist tends to fall apart due to repetitive slamming but the authors seem to believe they can do away with physical forces.

Let a thought experiment have a 20 kg and a 0,02 kg weight fall on their head wearing a sturdy helmet and let that thought experiment note the difference in their facial expression after that fall? Physical weight not relevant for design consideration of a helmet either, right? And you wonder why Planet RD and Planet AA remain totally disjunct.

The paper fails to mention all of these considerations.

Pulling also is different whether a flat smooth surface is pulled, a fabric or flexible cover is pulled, or whether a rope is pulled, a carton or box is lifted, and there, grips are different depending on the box handle, insert or whole.

The important role and relevance of physical force dimensions, loads, torque, pull or grip strength extents on sensible design and how it is constantly ignored by R&D

The role of load on design is very important.

Small or minimal loads may require a different approach to gripping than large or even massive loads.

When I distinguish between providing the overall shape of a power or cylindric grip for a 2 kg and a 40 kg item, the 2kg item may possibly be gripped by the prosthetic arm in the cozy and smallish environment of the split hook itself. A 40 kg item either will be bear hugged (such as, say, a contained robust 40 kg cardboard contained package), or a strong bag fabric will be wrapped around the hook tips.

Now, neither of these has been considered by above mentioned taxonomists - and why?

  • First and foremost, they start with a weaklish hook prosthesis to begin with;
  • secondly, they do not create a realistic scenario and let the amputee wearing a weaklish prosthetic arm fail that grasp activity, but the amputee works within a self-imposed restriction of work / life space.
  • Thirdly, they ignore the massive influence that weight, grip force and other absolute values in terms of loading must, should and will have on the specific design of the prosthetic components.

There are a few actual situations where force extent has played a massive role in prosthetic design for me:

  • Pull forces in my ADL by far exceed what a small spring operated wrist by Otto Bock could sustain. We built a steel ball lock wrist. This design was informed by actual use force parameters and tolerances, not by taxonomy. High pull forces are a direct consequence of applying high grip force operated split hooks in instances of lifting or pulling with large force.
  • Repetitive hard typing impacts onto the prosthetic hook tip would wear out epoxy socket screw holes that attach the wrist. Carbon fiber socket material was chosen instead. Also here, design had to be informed by considering force strength.
  • Grip strength positively correlates with cable pull strength. Conventional cable curve deflection sheath mounts would wear down and cause cable shredding after 4-10 days over two years or so. Replacing this design with a true Bowden sheath mount over a curved body surface remedied this, allowing for even higher grip forces. Very high grip forces can overplay restricted grip shape design, and is more versatile for the high end user with minimal degrees of freedom for prosthetic control. The problem of crappy sheath mount is identifed faster under higher loads - but the user dissatisfaction is the same with lower forces, only that the prosthesis does not crap out "already" after 4-10 days but "only" after a few months or so.
  • Covers tear if the motors are weak, and they are weak because they have to be light. The combination of a cover and a grip that both do not hold up leads to the prosthesis being discarded. Therefore, real forces are an absolute key to inform better design.
  • Making a prosthetic arm fail is a relevant step in prosthetic arm design improvement. Why? Because making the product (generally) fail is a relevant step in any consumer product design improvement. Interestingly, you cannot understand if you have not done it.

Injury risk and physical load dimension

The way you go about weights and loads directly impacts the way your body responds. Nothing impacts injury risk more than physical dimension of loads.

Not that prosthetic designers know that.


Hand of man - different physical force dimensions, different design

The Hand of Man shoiws in great clarity just how large forces enforce a totally different robot hand design. Watch and reflect.

(C) Copyright

More fun to read:

Attempts to describe the current situation in relation to prosthetic device / hand research: Valero et al.


"We argue that a truly bio-inspired approach suffers, by definition, from both gaps in our understanding of the biology, and technical challenges in mimicking (what we understand of) biological sensors, motors and controllers. Although biomimicry is often not the ultimate goal in robotics in general, it is relevant for humanoids and prostheses. Similarly, why is our understanding of the nature, function and rehabilitation of biological arms and hands incomplete?"
  • The authors seem to not respect the fact that good engineering first and foremost solves a constructive or functional problem. A prosthetic gripper needs to provide direct fast reliable control, reliable grasps, and passive characteristics that fit the user optimally. It seems as if these authors are lost in a different world. Yes, we can try to deconstruct and rebuild a full human hand - but what would be the point?

"Jacob Benignus Winsløw Jacques-Bénigne Winslow, (1669—1760) noted in his

Exposition  anatomique de la structure du corps humain  (1732) that ‘The coordination of the muscles of the live hand will  never be understood’ [33]. Interestingly, he is still mostly correct."
So given that these authors cite and agree with a guy from the 17th century, when he says that no one will ever understand the human hand, why are they totally oblivious to the fact that the engineering answer to this - also from a view centuries ago, the prosthetic split hook - also is correct?
What was wrong with agreeing to the fact that more than one obvious answer can be centuries old and not wrong and thus ** still ** be valid?
Who put the bug in all those people's heads that all of a sudden they can "re-build" a "hand", then cite a guy from the 17th century about just how out of reach a human hand is engineering wise, and yet, that now is a great time to look down on prosthetic hooks?

Real life taxonomy

Characterising the arm amputee from a control and equipment view

Interface control considerations

The biological human arm with hand has a arm-hand interface that is absolutely massive.

After my arm amputation I realized that the neurological vast real estate that once made up the control of my extremely capable right hand was, by and large, in progress of atrophy, lost, and gone.

The stump skin has only a fraction of the nerve density of the hand, and while the stump is very painful most of the time, its sensory use is extremely limited.

That means, that any device attached to the arm will have to work with an equally reduced interface width if it wants to conform to a paradigm of living life without visible means of support (see above). Myoelectric arms usually have only an open/close function simply because a good SNR (signal to noise ratio) is not available for more degrees of freedom. And really, my stump muscles are so massively atrophied that I would be shocked if any reliable signal came out of more than two locations there. Multi-electrode control may be possible "as is" but will probably have to be preceded by targeted muscle reinnervation - and that, in itself, was said to be error prone because the socket and interface can not always be reliably fitted to the stump so the electrodes are always where they needed to be.  So while any multi-electrode solution seems to be extremely error prone, experimental and thus full of more drama, one has to accept that after elbow amputation, including cost and time and efforts, degrees of freedom are now significantly reduced and that's it.

From that view, body powered control in the sense of using a cable powered arm seems like a scale-able, extremely direct and lightweight approach to the issue. Myoelectric control is not "thought" but "forearm stump muscle" controlled and thus equally body powered - any limb positioning effect, or sweat interference, will illustrate to you that myoelectric control is just as much subject to "body power" or "body unpower" than cable controlled system - only far more indirect, far more expensive, far more error prone and far more fragile.

Once you realize that - swallowing that may take a while, I know - you can then start to embody the bodily gymnastics that are necessary to handle myoelectric limb positioning interference, to reduce the impact of extended or lifted elbows on the gripper control, and you can improve the way myoelectric control works for you in a better way.

At one iLimb workshop, one Touch Bionics employee removed the prosthesis from her arm after the workshop and stated that her stump now was really sore. This is also a correlate of "body power".

And interestingly, with all the different works, my stump is never really sore wearing a body powered arm performing far more physically strenuous activities than what we did on that workshop.

So you ponder of the the ramifications of "body powered" being a term that explains the direct mode of cable control as much as the indirect mode of myoelectric control.

Direct body powered cable control brings about direct force feedback without time delay, far better lighter and far more robust constructions as well as mechanic rather than electric devices with far more robust characteristics in terms of cleaning and hygiene.

Data and control domain space considerations

The massive reduction of the interface width of the arm stump, compared to the actual massive width that a biological hand will need for optimal control, calls for statistical methods such as PCA (principal component analysis), as well as for a willful or arbitrary selection of grasps or functions based on relevance.

Time, with associated industrial product evolution cycles, has approximated that in the form of a prosthetic hook as offered by Hosmer in a few shapes, and the Becker hand, an invention that was first published in the USA and then improved over many decades to its current form. These products will embody the result of a functional PCA or reduction of dimensions to useful instances and variations.

So to inform better design can be done but it has to be done well to work in real life. Let us see how it may be done from a user view such as mine.

Current best solution and current best game changers for informed better design

Even without even looking at any new data, there are problems to be solved, prostheses to be improved, in the sense of a cold start:

  • The single most important game changer for the real world use are padded surface additions. Silicone tubes or gloves. They widen the grip performance while keeping cable load low. They reduce the need for a different gripper.
  • The second most important game changer then are extremely stable quick lock wrists that make for fast swap between various task specific hook shapes. With that, various hard two claw (or other) shaped grippers can be used.
  • The third game changer is the improvement of cable mount. There, forces that are usually limited can be jacked up while maintaining cable sheath integrity over a long period of time; usually, work relevant pinch forces and grip frequencies produce cable shredding with 4-10 day break intervals when employing regular or average, typical commercial parts.

There are things I do not perform, nor does anyone else perform these, because prostheses suck there. These are visionary improvement goals, however, they go where no one went before:

  • Bear hugs and large object transport. My prosthetic socket needs vulcanized or rubber-covered inner surface parts for that. I need these pads to be easy to swap, too. They will wear down like nothing else.
  • Combined grips, multiple grippers. You know how a gripper either provides a power grip, or, a lateral grip? And instead of switching these, can a hand or gripper provide multiple grips at once?
  • Specific item grip locks. There is a list of items we always use and need to grip the correct way well. One is using a knife to cut steaks. Another one is using the car steering wheel, where a lateral grip must allow swivel. A third grip is containing wallet, cell phone and other valuables with one encircling, firm, safe and entirely reliable adaptive grip.
  • Switching VO/VC/lock controls. Sure, the Toughware Equilux currently leads the pack when it comes to real work and real situation usability, with great functional performance, but can that be improved further? Can there be a lock at any point during the action, and can that lock be two-way? Sure, the Sure-Lok system of TRS stops the cable and thus gripper in a close position, but can there be stops in between?

The last aspect is the object. Objects can change, be changed, to suit the user. This is typically forgotten totally and entirely. But: I have a soft steering wheel cover, I wear Lock Laces, and so on. And people that once tried to find out whether this blog site was commercial or informative decided that it was commercial because I just had blogged about Lock Laces. It is as if they had no freaking idea either, that the lifestyle of an arm amputee that does go down the path of improving life with the "prosthetic option" in a wider sense also is a total material object onslaught. You buy stuff. You test stuff.

Statistics for real life use as an example to actually and really inform better design than design we really already have

As it is not possible to inform "better" design in empty space, it is is possible once one grasps the concept of "better" as "better than:..", as a relative term. Really, "better" can only be used, logically, in a comparison either against another device or against any normatively placed standard.

Comparing performance across devices and ADL and heavier tasks for real work

I logged a subjective performance (0-10, with 10 being nothing left to desire) for grasp performances of me and my various prosthetic devices, and, very importantly, of me without a device on. I then shoved these numbers through statistical software.

This was correlated with object sizes, weights and various other aspects. Among other aspects, a flag was used for activities where no failure for bimanual activities was a must (red data points in plots). As there are activities where one, for example, has a heavy very expensive camera or one drives a car, there, grips succumb to a no fail criterium. For all those, 80 or 90% success rates are still pure garbage.

Diagrams to correlate grip parameters in various use functions and grip performance across devices

The two top rows of the diagrams below show GLMs (general linear models) where I fitted object weights, sizes and other criteria - fragility, soft- or hardness - to model the various devices' performances. In a nutshell, the objects' physical characteristics also differently influence the device performance. So, grip performance is not an independent stable dimension that floats, by itself, in space and can be understood or grasped (bwahaha) "as is".

The third row models how non-relevant grips (grips one does not really need a prosthetic device for: black data points) and bimanual no-fail relevant grips (grips where basically your life depends on these grips not failing: red data points) compare within the different devices.

As it becomes quite clear, the Toughware Equilux has overall excellent grip quality but where it totally outshines any other devices is that it over-performs in the domain of bimanual activities that contain a no-failure tag. It is basically the best of the life savers there is. Compare this to the iLimb, that excels in grips that are basically non-relevant, in other words activities no one would need a prosthetic device to begin with and you will understand why advertising for these devices often contains the subtitle that "it is the little things that count". What definitely is a surprise, if anything, is that the stump (wearing no prosthesis) only slightly (and not massively) under-performs in the bimanual activities where failure is not an option, and far less so than an iLimb. Given that it is the cost-free option to not wear a prosthetic arm at all, and give or take a bit of practice, not wearing a prosthetic arm (again: you DID read the title of this blog, right?) is a totally valid option that, taking all factors into account, clearly may outperform any conventionally built prosthetic arm.

So the bottom row diagrams show two interesting rectangles that I highlighted with dashed lines: the left top dashed rectangle (low values X-axis, high values Y-axis) shows activities where the particular prosthetic device is just so much better than the stump alone. Those activities are relevant to the user and for those that look for selling points. And the right bottom shows activities where the stump outperforms the prosthetic device - and while embarrassing for sure if at all you are into that type of narcissism, it is also a great resource to study these activities if one wants to improve prosthetic devices to get everyone, users and non-users alike, to embrace prosthetic arms, and if only a few people at a time.

And for that, focus on activities of a non-failure category for bimanual activities could be a priority because those are the activities no one wants to see fail. That is how I would slowly start to inform designs. As it seems that the type of diagram that one makes to inform better designs nudges into the center of some recent research, I would put to these researchers that a focus on what really moves the amputee to using a prosthetic arm should dominate these diagrams. I will expect better, circular or polar, multi dimensional or otherwise convincing displays by any one that has the nerve for it - but so far, empathically and actively using my stump as base line has never failed to show the hard and clear advantages of a prosthetic arm over that base line, which, as stated, is, not wearing one.

So, grip performance of a prosthetic arm is relative, it depends on the actual object/s grasped or handled, and, it competes against the non-user employing Redarm Total Freestyle (TF), the absolute opposite of the Bluelight Special approach.

These diagrams, that understanding of a TF, are an absolute taboo for current research, something even the Makers of Cybathlon totally stayed away from, likely out of technical inability and lack of understanding: having a non prosthetic wearing arm amputee as comparison and have them solve any grasp problem their own way (TF!). And if they ever have to use their anatomical "normal" hand, so what?

My ability to use my left (anatomical, "normal") hand is exactly what places the requirements for the right prosthetic gripper into a different domain, given that it is what it is: an aid. It is that particular aspect that some 10-15% of Planet AA inhabitants think about permanently: will the arm be necessary today? My left hand grasps, currently, with a closing force of maximally around 47 kg. That could be higher, but when handling bodies of 60-90 kgs of weight on average, it seems useful.

Specifically, there are a few activities where a prosthetic arm needs serious trouble shooting to make it the better option than not wearing one:

  • Driving car. While the prosthetic arms are not all that bad, doing that just with the arm stump is quite simply better. The problem is that a prosthetic device with a circular type grip - iLimb, Becker hand, Hosmer 6, all Trautman derived designs - risk to enclose the grip around the steering wheel, locking the socket into a particular angle. The only grip I found to work well is a flat compression grip, that stabilizes a softly padded steering wheel and allows a wheel swivel while maintaining grip. A Hosmer 5 steel hook is the device to wear, for sure.
  • Handling wallet and keys, as well as cell phone, like, at once. While that seems strange, the issue is that the situations where one does hold these objects in one hand are often those that contain comprehensive body motion. Such as sitting down in a train. Or getting in or out of a car. Full body posture change is a situation that makes myoelectric arms act up, politely put. The best option so far was to avoid these activities altogether: do not handle the wallet at all but leave it placed in a bag. Do not handle the cell phone altogether but leave it in the pocket. However, I use the mobile phone as navigation system so there is that. And while the grip performance of various prosthetic gripper devices are not all that bad, doing this with the arm stump alone is just better. And when I decide about wearing or not wearing the prosthesis, and all I will use it for is driving and getting in and out of the car while holding my wallet and cell phone, then the prosthesis stays at home, unused, and that is the decision you would not want me to make.
  • Transfer a (human) body from one place (i.e., striker, gourney) to another (such as a CT-table). Also, that is relatively heavy work. It works really well using the stump, to push, and to push with care and feeling and no fear of damaging the prosthesis. Obviously, it is both a strenuous and gentle task: the body is not to be injured or damaged at all.
  • Cocktail party. While that is not on the chart here, not wearing the prosthesis for that also means that it cannot fail. And cocktail and similar parties where one stands, gets around, talks with people while handling food and drinks, are to 100% no-fail events. You can do anything there but you must not drop an item. Ever. And while the arm stump is not the greatest asset for gripping plates or drinks at a cocktail party, any prosthetic gripper is so slightly problematic that it is difficult to decide for wearing one, as if the risk of failure is anywhere above 0%, better to not wing it and stay safe. There, a Becker hand or Toughware Equilux would be my preference. Or, playing it safe and not wearing a device at all.
  • Opening a laptop or a watch for repair. Or: replace or mount ceiling device and electrically connect to the copper wires there. For these activities you need subtle manual abilities plus, on any occasion during that type of work, a very precise and very, very hard push or precision grip. The only device I know that does that really well is the Toughware Retro. And that was not tested in my currently analyzed lineup of devices. Other than that, the stump, plus a few tools, are the way to go.
  • Writing. There is only one hook device that I have that is specifically built to hold a pencil. To be honest, it is one of the activities that feel most civilized - getting my prosthesis the ability to write and perform brush strokes. And  did build a guitar pick / plectrum mount for my Hosmer 5 hook.
  • Taking samples is a tag I used for holding a test tube and managing the manipulation of a cotton swab. The issue there is the same as with flipchart pens: there is a firm lid or cover connector, and when one opens or closes it, the grip of the test tube needs to be extremely firm without, however, damaging it. So far, I found free-styling this the best way, until the Toughware Equilux came along.

The aspect of weight in context of grip performance across devices

Does weight of the object really play a role (diagram below / next diagram) ?


And if you want to understand how to build and improve prosthetic devices, then there are relevant differences.

When examining only those activities where bimanual no-failure activities are relevant, the difference to performing with a stump can be obtained, here, from grip subjective performance ratings. If you find objective ratings for grip performance, great, you could go and analyze research video footage maybe and get your name out - but for time being, subjective rating seems to be perfectly sufficient for rock'n'roll, which is what this here is. So: a performance that is better than a stump performance by 5 points would rate as 5 on the y-axis on below diagrams.

What we see below thus are three devices: iLimb (left diagram), Hosmer model 5 standard steel hook (middle) and a Toughware Equilux (rightmost). There are two interesting numbers or quantifiers: of all activities that do not allow for grip failure (11), how many does the device perform better than when using the stump, and how many are worse when using the device?  The upper number gives the first, the lower the second count. And, what are these? See the annotations.

As we see, an iLimb has performed better than me just with the stump on two accounts out of eleven, and it has performed worse on seven accounts out of eleven. If that does not motivate you to accept that stump performance, working situations without prosthesis on, is the best base line ever to see where prostheses become a problem as sold, then I do not know what. Maybe come over so we can discuss this 1:1.

A Hosmer model 5 hook performs better in 7 and worse in 5 situations whereas the Toughware Equilux is better in 7 and worse in 0 situations. No wonder the Toughware Equilux does make my day, as I get to use up a lot less stress hormones over a few days of real work. That difference is not present across various devices for non-bimanual activities or for those where it simply does not matter whether the grip fails or so. There are such situations, such as reading newspaper. There, no sanctions are imposed on failing a grip. And, so, no, it is not the little things that count. Not here.

In relation to object weight, the iLimb fails all higher physical load grips, also because it is weak and has a crappy glove or protective cover. While it is no surprise, it is a clear fact.

The reason why the Hosmer 5 hook fails the cell phone and wallet holding situations is because of a geometric relation between grip shape, hook material and the materials of the objects being held where grip force is too low to make up for the crappy object-hook form closure.

The Toughware Equilux has both a better gripper shape and surface option (which I totally exploited), and a better control paradigm, so it can perform a lot better in the domains where the grip forces of the other devices are too low to make up for their geometrical failure.

To see how the grip performance compares or declines over various object weights, it is best to see these side by side (diagram below). There, object categories are 1: <0,1 kg 2:<1 kg, 3<10 kg and 4 the rest (obviously heavier).

There, it becomes apparent that - while complaining on high level - the VO (voluntary opening) control of the standard hooks is less ideal for class 4 weight range objects unless the gripper back-locks (Becker Lock Grip and the same mechanism in the Becker Imperial hand, and I sure forgot to rate my tweaked back-locking Hosmer 6 model here) or unless it can be controlled via a VC (voluntary closing) mode (Toughware Equilux, TRS Prehensor).

Practical people like me regard the design modification of switchable VO / VC over, say, myoelectric control, as absolutely massive - and your mileage may vary.

Also I will see a back-locking mechanism as having fundamental advantages over a non back-locking gripper, and you would only concur had you ever to operate a sharp hedge cutter hanging on such a gripping mechanism, as otherwise that design spec would be countered by a deer in the headlights look in the face.

Weight plays a role just as far as the grip quality is not sufficiently provided by material shape deformation and geometry details of the gripper.

However, total weight is just a crude approach to evaluating designs for better information.

For example, opening and closing the prosthesis, regardless of object weight, repetitively will take a serious toll on the cable and cable sheath, even if the object is lighter than 500 g as an example. Lifting past 30 kgs of weight several times a day does so a lot more, particularly when using a VC system. The design of a prosthetic arm or its failure directly links shape, geometry, material choice, technical design and allowed loads. Thus, the defining characteristics cannot be sensibly separated. Heavy typing will create significant back-and-forth motion of the prosthetic socket against the forearm skin if that socket is not suspended on the forearm really well.

Only a very well designed fit can make sure the prosthetic device stays in place and the forearm stump skin is not quickly abraded. The whole application is tightly coupled to the actual weight of the distal parts, i.e., wrist connector and hook. Obviously the back-and-forth motion conveyed to the socket by the typing motion is dependent on the weight of the distal parts. With a flat elegant type access angle with a Hosmer 5 hook, shoulder elevation for typing is minimal and horizontal acceleration of the socket is not that high. Different geometries result in slightly different accelerations however. Overall, one must understand these interrelated aspects between shape and applied forces before being able to understand actual problems or being able to provide actual solutions.

PCA of grip performances across devices

Qualitatively, working out subtle performance aspects can be difficult for all of the devices that I looked at. It is not immediately apparent why I need a few of these in my bag if I go to work or on holidays. Now, the subjective performance ratings do yield different vectors in the PCA (diagram below). This shows that having an assortment of devices other than an iLimb allows to cover a rather broad range of performances in a well differentiated way. People that have no idea about using prosthetic arm devices will put all body powered devices into the "old rubbish / Captain hook" category, such as the organizers of Cybathlon do - forgoing any relevant application of actual rehabilitation devices that work under real work constraints. One not so subtle interpretation of some other dimensions I had plotted there is that many prosthetic devices particularly fail at very small and thus very low weight objects - while mastering the upper range by use of TRS Prehensor, Toughware Equilux or even a Becker Hand is no problem, either practically nor financially. The placement of the data in the PCA diagram also allows one interesting interpretation in that functionally, coming from voluntary opening (VO) hooks ("conventional hooks"), the voluntary closing (VC) devices both reside on the "other side" of both prosthetic hands in the test. That suggests that these VC devices exaggerate whatever it is that the hands do, technically, which, mostly, is both offering an adaptive grip and a deformable cover. The fact that a VC prehensor may be better than prosthetic hands given typical everyday situations is also illustrated by the fact that a TRS prehensor type device was the only device to complete all points at the 2016 Cybathlon, and by the fact that due to that, it won. However, given work situations not included at this 2016 Cybathlon, a Hosmer hook may in fact be the better choice.

Grip taxonomy in context of performance

From a practical view, therefore, classifying grippers in context of actual grip situations, particularly after a performance seeking user as myself has gone over his devices and already added modifications, is not simple or easy. There, the devil is very much in the details. A technically sensible approach to modifying prosthetic device designs further in order to inform better design contains the inclusion of sliding or overlapping categories, subtle differences and inclusion of rather differentiated design modifications, object dimensions and weights. The conventional grip types are not entirely irrelevant. One can use them to compare where a prosthetic device performs better than when one does stuff without prosthesis. While the inventors of grip type analysis may not have considered that, it helps understand how arm amputees may find a prosthetic arm not just as useful as the commonly believed advertising would promote this. The next diagram shows the relative performance, versus not wearing a prosthetic arm, of a Hosmer 5 hook (top), of a Toughware Equilux (middle) and of an iLimb (bottom). One trick to better understanding life is indeed to provide cogent grip patterns rather than general or non sequitur discrimination attempts as seen elsewhere. In particular, holding a steak knife to cut food when eating / dining will require a particular grip that I called "power cut". You could also call it something else if you do not agree. Neither the iLimb nor the Toughware Equilux really allow for that grip out of the box. As the Toughware Equilux has great potential otherwise, I custom built soft gripper inlays or pads that boost that grip type and remedy the problem. In order to exceed any push capabilities of my stump (right below elbow amputee - hence the name of my blog in case you forgot), a prosthetic device probably needs to be really really good. The reason a Hosmer or Toughware Equilux device rates less than my stump for push functions is probably owed to an unweighted averaging of all the single different activities that I do. If this was weighted for vibration, frequency and duration, as in heavy keyboard or typewriter typing, then a Hosmer 5 hook certainly would win over many other options. When one compares the way my right elbow's "power" grip performs against an iLimb's "power" grip, the difference is not just as much in favor as some hoped and yet, it may explain why not everyone feels it is so absolutely necessary to shell out 80 000 bucks for that puppy given its comparative yield.

Also, we can compare small device differences across grip types following a sort of "taxonomy" as suggested by the papers reviewed above. A Hosmer 6 hook outperforms a Hosmer 5 hook in the pinch / precision grip, power grip and push domain. This is certainly owed to its different shape as all other aspects are same. The TRS Prehensor with its entirely different control paradigm does not align as neatly with the Hosmer 5 hook: its vast improvements do not become apparent with grip type comparison but (see above) only when considering grip function or degradation across increasing object weights. Conversely, it is the VO/VC flip option as well as the different shape that elevates the Toughware Equilux over the TRS Prehensor even though other average comparisons (see weight related charts above) do not show just a large difference between the two.

By adding several relevant rather than irrelevant sub-types for grip categories or grips - such as a specific driving steering wheel grasp type grip, or a specific keyboard typing push - I can create different and a lot more specific diagrams for grip categories. If one wants to categorize at all, then that is a way to go: include * meaningful * categories. These very necessarily contain specific weights or loads, physical requirements and with that, design requirements, in order to make sense.

General considerations at the end

A direct technical appraisal will clearly show that specific problems need to be weighted and put into context before they are successfully solved in what clearly is a very narrow and highly constrained space.

With that, the question really is how to weigh aspects, what aspects to weigh, and how to provide generalized solutions for specific problems.

From that view or angle, copying the "design" of a human hand (which requires a few thousand single controls and sensors) will make you require or supplement a few thousand single controls and sensors, either by going gaga with the I/O real estate, or by programming some simulation and coordination programs, which is why that design type, to cut it short, does not appear greatly useful. It can be made to work to some degree, but you lose along that effort.

You really want to keep efforts sleek and close to the realities of the constrained application in order to approximate a maximally lightweight and useful device. The option to swap terminal devices contains the option to omit or vary functions across these, making them both specific and light weight - an ideal aspect for prosthetic arms.

Grip wise, the following grips will require a particular focus:

  1. Sliding swivel grip for steering wheel. Flat and moderately strong only. Definitely voluntary opening. Hosmer 5 hooks have that.
  2. Soft adaptive very secure grip to hold valuables such as cell phone and wallet, or a camera, securely. Definitely voluntary opening. The Becker Mechanical hand has that.
  3. Soft adaptive very secure grip for all types of kitchen work, voluntary closing. Toughware Equilux has that.
  4. Push surface. Check V2P prehensor, for an ideal push design.
  5. Special grip for cutting knife - kitchen work, cutting bread or slicing vegetables or meat, eating pizza or meat. The Hosmer 5 is quite good but not perfect here (larger knives are an issue).
  6. Grabbing, across power grip to lateral grip (adaptive shape of grippers), with force feedback power, definitely voluntary closing, for dynamic manipulations. Toughware Equilux or TRS Prehensor are good there.
  7. Precision grip, and I mean a "precise precision grip", where the planes of the grippers predictably grasp. Required for all work bench, sewing and similar manual work. Hosmer 5 hooks have that, it is great.
  8. Carrying grip for all kinds of handle containing baskets or bags, including holding on to a heavy machine handle bar reliably or a vacuum cleaner tube or water hose. Lock grip, possibly following Trautman hook shape, but the Becker Lock Grip or Imperial hands also have that.
  9. Lifting heavy bags onto the shoulder by their strap, lifting boxes: round curved outside for one, rectangular for the other. And pretty stiff to hold up. Look at the Hosmer 5 hook design.
  10. Typing may be performed on an industrial level. Not all wiggly parts survive that. The right shape of the hook or typing device also may end up deciding over the amount and extent of shoulder, neck and back pain after the typing. So if you have 4D coordinate measurement systems, here is where to use it for optimization. Otherwise start with a Hosmer 5 hook as design template.

Must have as part of active / actuated gripper mechanism are 1 and 4. Passive pressure against objects against belly / body is possible for 2 (stump against body) and pressure against objects on surface (such as kiitchen table) is possible for 3, using amputated arm's elbow for 5 and 6 works well, using the stump for 7 is not a problem at all. And yet: if a prosthesis screws up on a relevant situation across 1-7, it may become a real nuisance, and one ends up not wearing the prosthesis at all, supplementing all 1-7 with tricks and no prosthesis on.

As examples: a prosthesis that works well when working on a work bench but sucks for driving will impede me critically for a critical step (driving) so I am better off not wearing it there and then being inventive with work steps at the bench without prosthesis on. The prosthesis may be somewhat cool but when it is too heavy so I cannot type heavily for 5 days a week,, then I will have shoulder pain and the arm won't get used.

Repairs as design aspect to consider and possibly improve

I just broke my steel cable during holidays when biking, after I definitely had overused my prosthesis with very extensive and heavy lifting and works, and fixed it with scissors, a bit of tape and a Kevlar string (YELLOW, above image, tennis racket replacement parts, can easily be ordered quickly from many online stores). If I still have to tell you just how elegant, how fast, and just how relevant, just how important that is, now, as a prosthetic setup for serious mountain biking in particular and for work rehabilitation in general, then let me know. If you do understand this in its whole scope and meaning and what exactly it means for you or anyone else trying to build prosthetic arms, well, then: Cheers.


To inform better device design, the design deficiencies of current designs in context of the actual use are one thing (and may mostly pertain to creative off-label variation use or reliability testing), but the design deficiencies in context of intended, necessary, required, planned, hoped for or enabled new uses weighed by importance, and weighed by orthopedic long term impact, are how really better design is informed.

There, grip type performance, grip type needs and actual repetition, weight and torque aspects all play into the question of whether the device really succeeds on the arm of an arm amputee or not. And that decides the question whether a particular design is, in fact, "better" or not.

[1] R. E. Stepp and R. S. Michalski, "Conceptual Clustering: Inventing Goal-Oriented Classification of Structured Objects," in Machine Learning: An Artificial Intelligence Approach, 1986, pp. 471-498.
    author = {Robert E. Stepp and Ryszard S. Michalski},
    title = {Conceptual Clustering: Inventing Goal-Oriented Classification of Structured Objects},
editor={Anderson, J.R. and Michalski, R.S. and Carbonell, J.G. and Mitchell, T.M.},
    booktitle = {Machine Learning: An Artificial Intelligence Approach},
    year = {1986},
    pages = {471--498},
    publisher = {Tioga Publishing Company}
[2] D. L. Medin, W. D. Wattenmaker, and R. S. Michalski, "Constraints and preferences in inductive learning: An experimental study of human and machine performance," Cognitive Science, vol. 11, iss. 3, pp. 299-339, 1987.
  title={Constraints and preferences in inductive learning: An experimental study of human and machine performance},
  author={Medin, Douglas L and Wattenmaker, William D and Michalski, Ryszard S},
  journal={Cognitive Science},
  publisher={Wiley Online Library}
[3] A. J. Spiers, L. Resnik, and A. M. Dollar, "Analyzing at-home prosthesis use in unilateral upper-limb amputees to inform treatment & device design," in Rehabilitation Robotics (ICORR), 2017 International Conference on, 2017, pp. 1273-1280.
  title={Analyzing at-home prosthesis use in unilateral upper-limb amputees to inform treatment \& device design},
  author={Spiers, Adam J and Resnik, Linda and Dollar, Aaron M},
  booktitle={Rehabilitation Robotics (ICORR), 2017 International Conference on},
[4] J. T. Belter, B. C. Reynolds, and A. M. Dollar, "Grasp and force based taxonomy of split-hook prosthetic terminal devices," in Engineering in Medicine and Biology Society (EMBC), 2014 36th Annual International Conference of the IEEE, 2014, pp. 6613-6618.
  title={Grasp and force based taxonomy of split-hook prosthetic terminal devices},
  author={Belter, Joseph T and Reynolds, Bo C and Dollar, Aaron M},
  booktitle={Engineering in Medicine and Biology Society (EMBC), 2014 36th Annual International Conference of the IEEE},

Cite this article:
Wolf Schweitzer: swisswuff.ch - Understanding and informing design issues of a prosthetic arm for below elbow amputation by way of "taxonomy" [literature review, reality check]; published 26/07/2018, 21:18; URL: https://www.swisswuff.ch/tech/?p=7651.

BibTeX: @MISC{schweitzer_wolf_1653036324, author = {Wolf Schweitzer}, title = {{swisswuff.ch - Understanding and informing design issues of a prosthetic arm for below elbow amputation by way of "taxonomy" [literature review, reality check]}}, month = {July}, year = {2018}, url = {https://www.swisswuff.ch/tech/?p=7651} }