Towards Extreme Cyborging (EC) microworks: very small things [grip mechanics theory, parametrization and then testing side by side - Trautman hook, Hosmer Mod 5 hook, TRS Prehensor, Touch Bionics iLimb revolution]

Small things to be picked up, side by side demo with other prosthetic devices.

Yes, very small things.

"It is the little things that count", they said. "The little things are important", they said. They all said that. But behold, their "bionic" apparatus cannot handle little things! "Why is a hook not evil", they wondered. And a storm of little demo videos came upon them. Deep into the myriad of grip mechanics this went.

"Get a grip on grips", he said.

This totally bypasses the fact that after laying down real life consequences for what I call Extreme Cyborging, I just finished building my first own steel Trautman hook, and all just with shape information from the internet. Yeah, you read that right. I did not build yet another one of the ubiquitous funny hands that promises to change my life or what it is these 3D printed hands now do - no. I sit on the demand and I sit on the technology and what is it that I do? See? This is what should really disturb you - because given the current signs of the times, it should feel deeply wrong on many levels. To you. But then, we were likely living on different planets all along, and maybe it is time you realize that too ; ) After we knew since years in detail what the Cybathlon showed us yet again, with glory but not with any improvement, it is yet again up to us, the users, to push further into what is, what can be, and what matters. And honestly? We have seen far enough funny hand videos.

What we have not seen is the Rebirth of The Cool. The Rebirth of an absolutely insane gripper. Physically. In 3D printed steel. The Trautman hook is such a device. And I went for it just because I can.


Not everything that moves also grips.

Grips can fail, and it does not take a degree in brain surgery or rocket science to know that.

Grips, to be grips though, need at least two gripper surfaces X and X' to absolutely minimize errors E to underscore some ADL-related epsilons, with at least some five ranges of error:

  • (L) approach each other so that regarding locations L, L(X) - L(X') approximates zero, possibly with a remaining distance of E(L);
  • (A) provide parallel angles of possibly planar if not interlocking surfaces so that regarding angles A, A(X)-A(X') approximates zero; possibly with a remaining angle deviation of E(A);
  • (D1) provide a shape match against each other so that given an approximated surface lock of X and X', the total of a minimal distance map D1(X,X') approximates zero; remaining distance then E(D1);
  • (D2) provide a shape match against any planar hard surface if objects need to be picked up, so that the total of a minimal distance map D2 for the gripper operating directly on a hard surface H described as D2(H, X, X') approximates zero, with a remaining error in E(D2);
  • (S) provide a stable configuration under strain, so that with increasing levels of strain on the prosthetic gripper, E<-S with E(L, A, D1, D2) being a function of S, remains minimal over increasing S; whereas S may quantify mechnical, thermal or other strain;

With that, any object size O and the metric absolutes of E directly relate.

There are a few other dimensions that come into play, over hours, days, weeks and months of use, for a more practical rating of any type of prosthetic arm device (or for not wearing any) - like, integral under the operating power curve, center of gravity, break down / failure rate, gripper friction and deformability, as well as a few application specific aspects that are not of a general nature. A really comprehensive rating is the End Of Day Feeling (EODF).

With that, you can now easily formalize any observations that you make further down here (or elsewhere on this web page).

If you associated ADL-related activities with relevant thresholds e (epsilons) for above mentioned grip descriptors, below mentioned figures will automatically tell you what device is best.

Dry exercise in rating grip parameters

Device L A D1 D2 S
Hosmer Mod 5 hook (steel) Very precise location of tips (steel, ball bearing). E(L) < 0.1mm. Very precise angles of tips (steel, ball bearing). E(A) near zero. Very precise shape match, easily holds planar paper (80gpsqm) with >2kg. Slanted shape makes manual approximation to achieve a E(D2) near zero a bit of a visual challenge. The device certainly ranges in a E(D2) that is below 0.5 mm. Steel, ball bearing, makes the devices endure high strains, thus clearly allowing for very low E(S) throughout the usual spectrum of S. Hooks are also a first choice for many farmers.
TRS Prehensor with user added soft tip covers Very precise location of tips (hardened steel). E(L) < 0.1mm. Precise angles of tips (hardened steel). E(A) near zero. Very precise shape match, containing a curved shape lock of opposed gripper surfaces. With soft grip cover add-ons, an extremely low E(D2) makes for relaxed and reliable pick-up of objects sized below 1mm. While the device contains plastic parts, it allows for massive grip forces due to the VC (voluntary closing) control. It is built for high levels of strain, so under increasing mechanical strains, E(S) is not known exceed limits.
Trautmann hook (rebuild) Relatively precise location of tips (overall design). E(L) < 0.5mm. Very precise angles of tips (steel, ball bearing). E(A) near zero. Very precise shape match, containing interlocking spiked strictly parallel surfaces. Orthogonal position to hook frame targets a E(D2) near zero without visual challenge. The device certainly ranges in a E(D2) that is very low and can be improved by better manufacturing. Steel 3D printing may not result on the highest tensile, mechanical or thermal stability. However, this device is absolutely great.
Touch Bionics iLimb Finger tips do not necessarily meet, index/thumb E(L) may exceed 1 cm. Finger tip angles uncontrolled, uncontrollable. E(A) wildly unmanaged. There is no shape match with fingers both containing rounded glove tips with contained sharply angular plastic structures. Thus, E(D1) may actually exceed 2 mm. With rounded gloves and unmanaged shape match, E(D2) exceeds 4-5 mm thus making this device useless for picking up very small objects from a surface - not by accident but by design. With very costly components even in part  disintegrating outside any use whatsoever, E(S) has to be assumed to be immeasurably high with regard to mechanical, thermal, chemical and non-use related strain alike.

It already is clear as sky, from the very outset, by simply performing a side-by-side comparison of the very simplest and obvious aspects of grip mechanics, just how the cards are dealt.

A good technical advisor, or prosthetic technician, and definitely a good mechanic, knows all that of course, and can tell you beforehand how any type of grip test will turn out, or at least, give useful indications.

So, here is to more closely observing the technical aspects of grips.

Actual performance of devices

Trautmann hook - EC microworks

Yeah, the rest of the world tries to top each other in printing 3D hands whose fingers do who knows what, but not have tips coordinate grips - and what do I do? I chill, and get myself a Trautman hook printed. In steel. Jeez, it even looks old directly after assembly.

But then, it grabs and lifts the very smallest rubber - and that is nothing to sneeze at. Not if you look for actual grip support, like, for your prosthetic arm.


Small rubber

Smallest rubber


TRS Prehensor - EC microworks

The TRS Prehensor is the perfect tool work device because it can apply as little or much grip force as ever you want to.

I modified its tips a bit and here it is, with the very smallest of things.

Detailed spring grasp documentation


Small rubber


Smallest rubber

Hosmer Model 5 hook - EC microworks

The Hosmer hooks of the model 5 variety (steel, aluminum, possibly with nitrile covered grippers or, as this one, just metal) are slanted.

Now, the slanted aspect gives the hook a performance advantage for picking up objects sized over 4-5 mm from a surface, as the slant manages to roll these off the surface (see Yenga test for a detailed documentation).

At the same time, this design makes it harder to visually check extremely small parts getting grabbed, and so the struggle here is in getting hard to see items lign up with the hook. The grip itself is perfect, obviously these hooks have been improved over decades and offer a great grip, nowadays.


Small rubber

Smallest rubber


Touch Bionics iLimb - EC microworks

The iLimb is a great "bionic" prosthetic hand that comes with a lot, and I mean a *lot*, of promise and expectation. That does not mean that it delivers - no. But it comes with a lot of hope.

So one has to be fair when testing it: it was built to be waved around, and thus its test should only consist in elegantly waving it around. Who poses with it on a bicycle, who poses with it with a vacuum cleaner, will thus prove they did not understand what an iLimb really is: *not* to be used. You cannot use it. It fails.

I already discussed symbolism in our contemporary society in a lot more detail elsewhere. With that, to use an iLimb for any real grip type application means to desecrate one of the really important symbols of human progress - which the iLimb undeniably wants to be. One cannot possibly take this pig to market, it is not built for that! At best, it manages a "gesture". Its role is to symbolize, to represent, to exude hope and belief, to "be" ("like"?) a "hand".

Even more, it clearly positions its wearer and owner into the "leisure class", and it really makes a very strong statement for the wearer and owner not being a worker, or blue collar type person. Only manual labor type guys pick up screws and parts, and who gets caught actually touching a real device loses this type of game.

To pick up small objects? The iLimb would have be wrongly constructed as symbol of anthropomorphic leisure class symbolism if it would now actually grab these items. And so in a reverse analysis, it is successful in that only if it really fails here.

As we see, there is just no way that you can get the small items picked up with this puppy. Success! No iLimb wearer has to ever fear others ask them to pick up tiny items.


Small rubber


Standardize tests as much as you want - and still find your prosthetic arm product tested with any type of ADL related activity

Standardized tests make sense for devices that undergo standard usage.

A prosthetic arm however is a cumbersome lifestyle enhancement with the optional orthopedic benefit to the user's asymmetry and overuse problems.

Its benefits usually are outweighed by the cost and effort, and many unilateral "bionic" arm users may in fact experience significant overuse of their other arm simply because, as can be seen here, the "bionic" arm quite simply does not perform, like, at all. And not for lack of trying. It just does not offer any grip advantages on a useful scale. That said, I love my iLimb to death for purely emotional reasons. It looks so much like a hand! Not that it performs like one ; ) It is more meant to be waved around, and with that, to "give life back" (whatever that means) and make one grateful "for the little things that count" (not that it actually handles even little things, bwahaha).

ADL (activity of daily living) tasks are notoriously demanding, difficult to classify and hard to even understand. Hell, no roboticist would have, out of the blue and upfront, asked an arm amputee to pick up any of the objects shown above - and my work bench is full of such stuff. With that, we all freestyle, and within freestyling, prosthetic arms may find a different place than within some imaginative fidgets of obscure test standard imaginations. Real life users will take all these devices to test them, immediately, without mercy, and with very good reasons.

So build your arm, to succeed in real life. What cannot be used to build a prosthetic arm that succeeds in your life may not want to be part of it.

The best place to find out about actual tests, test cases, use examples and hard problems for prosthetic arms is, therefore, with the users. It is with the people that, ideally, wear them at work and for activities.

Watch the grip

To identify grip issues, watch the grip.

Eyes on the grip.

Watch it.

It can be disconcerting. I know!

It can be painful to see how along the above mentioned variable dimensions, things derail.

And with that said, I can tell people that do not watch grips. Those also be the ones that hail iLimbs for what they are not.

Watch. The. Grip.

Cost function

Yeah, prosthetic device designers nowadays assume that we, amputees, exert some type of cost function.

They also assume that we do not operate in a linked, open world.

They operate under a closed world assumption.

Now, I have a few interesting cost functions of my own for my prosthetic arm.

  • The more items that fall down, that I break, that I let slip, the worse my standing, the worse I am off, the less likely I am also getting invited or admitted again. This may apply to, but not exclusively contain: friends, family, shops, malls. This is a complex sociofinancial cost function. Then, grip error rates have to be way below 1/1000 to 1/10000 not for laboratory settings but for the real life mess, with sweat, uncharged batteries and all.
  • The less I wear my prosthetic arm, the less repairs I need. The less repairs I need, the more I can do other stuff that I need, and, want, to do. It saves also money. And emotions. And energy.
  • The less I wear my prosthetic arm, the more proficient and thus reliable I get doing stuff without it on. See, prosthetic industries think it is a great money making feature to, as with the Touch Bionics iLimb, sell an edgy hand skeleton that rips apart gloves, then have gloves rip apart, then have the hand so weak that gloves must be thin and delicate as thicker gloves stop motors from working, then ream users with the gloves. I kill a glove within 15 minutes of very light use, or less, if it does not just fall apart all by itself. At the same time, the electrodes now started to cause eczema on the stump, possibly due to a strange metal-electricity-skin-heat type interaction. So I necessarily need to not wear a prosthetic arm for a greater part of the day, week, month, year. With that comes the unplanned "training" of doing stuff as is. After one has gathered that type of training, it is very hard to just go back to being fully dependent on a prosthetic arm.
  • The more I work to get my body powered arm set up as *extremely* robust, the more I can wear it for stuff that actually is hard repetitive bimanual work, and the less per year occurrences I run up, of all things related to asymmetry and overuse. The prosthetic arm that does, however, survive hard use for hard bimanual activities, is a body powered hook. The iLimb will faint all by itself in the cupboard, I am sure.
  • The more I wear my myoelectric "bionic" iLimb for sweaty activities such as bike riding or hikes, the more I get eczema on my stump, and the less I can wear even the body powered arm. So I cannot wear the electric stuff because it damages me.

When I stick to (1) not wearing the body powered prosthetic arm for all activities that I do not really need it for, and (2) wearing it for hard bimanual activities or work, then my cost functions are optimized. Grip reliability is best for the stump without anything on it, and, the body powered arm with one of these gripper or hook devices on it.




Cite this article:
Wolf Schweitzer: - Towards Extreme Cyborging (EC) microworks: very small things [grip mechanics theory, parametrization and then testing side by side - Trautman hook, Hosmer Mod 5 hook, TRS Prehensor, Touch Bionics iLimb revolution]; published 28/12/2016, 17:52; URL:

BibTeX: @MISC{schweitzer_wolf_1653033264, author = {Wolf Schweitzer}, title = {{ - Towards Extreme Cyborging (EC) microworks: very small things [grip mechanics theory, parametrization and then testing side by side - Trautman hook, Hosmer Mod 5 hook, TRS Prehensor, Touch Bionics iLimb revolution]}}, month = {December}, year = {2016}, url = {} }