A recent article [R. Baines, D. Shah, J. Marvel, J. Case, and A. Spielberg, “The need for reproducible research in soft robotics,” Nature Machine Intelligence, 2024] brought forward a list of interesting points discussing soft robotics research [1] ¹

Comment 1. Paper states “soft robotics technologies are scarce in society. This deployment gap is due largely to uncertainties surrounding the absence of standards, as well as difficulties in replicating published solutions”.

Evidence shows otherwise with regard to prosthetic arms. Prosthetic arms are a killer application, a go-to-solution, by concept however, not so much by product. Not a single aspect of reality here is anywhere near satisfactory, leave alone perfect. There are many, mostly useless, published solutions. No one cares a bit about standards. No one replicates a thing. And yet, prosthetic arms are an industry, they are a total thing. Everyone surfs on a different side of these ever rolling waves of prosthetic arm realities.

Possibly, soft robotics otherwise can learn from that setup, that plays various differing needs against each other.

Thereby, research does research. Industrial manufacturing does industrial manufacturing. Medical product definition does medical product definition. Medical prescription does medical prescription. Medical workplace adoption does medical workplace adoption. And user does user.

Research does research (academic). One may provide (or not) strict method research, such as identifying hard proper test procedures. There, prosthetic arm research has not made tangible progress, despite clear indications from the user view, such as here [e.g., paper, SHAP, appearance, hard work specifications]. Research also could, in theory, raise new interesting ideas. In the domain of prosthetic arms used for real work [see paper], contemporary projects have yielded absolutely nothing [EU], whereas for example also detail aspects such as myoelectric correct control rates have not progressed, but even slightly declined, over the years [e.g., 40 years of failure, non-understanding of use cases in Cybathlon], which results in extreme costs for users [cost]. You could list researchers, supervisors, funding bodies, reviewers, all – no one cares, no one acts. The academic research situation generally, in the domain  of prosthetic arms, has to be regarded as dire in various ways, including e.g. the mastering of simple anatomy [link] or just the understanding that sweat is a regular daily occurrence to be factored in before even starting to build prosthetic arm controls [e.g., sweat, boys, rash, dialog]. A very simple normal flow chart is not understood or adopted [flowchart]. And yet, research funding bodies [EU] cannot stop themselves throwing more money after what has to be absolutely useless research for actual rehabilitation purposes [junk].

Industrial manufacturing does industrial manufacturing. There, manufacturing, and technical quality controls, and legal compliance, are relevant. Industrial manufacturing in prosthetic arm domain has stalled and yielded non-satisfactory products, despite clear user indications such as from here [many but e.g. Otto Bock cable, Otto Bock wiggle hook, Otto Bock bolt/wrist, Touch Bionics / Ossur gloves, Fillauer steel cable add-on]. The customer service in this area has to be regarded as difficult or non-existent, virtually, from my direct experience. Compliance is a super difficult subject, there are a lot of open questions, yet to be examined with more scrutiny [CE].

Medical product definition does medical product definition. There, and medically, the goal of reducing overuse and asymmetry effects factually is a first priority, whereas no current industrial or research based focus or approach seems to use that to inform any design to inform any build, or to even inform any project path [overuse]. It does not help that manufacturers seem to not always consider medical product requirements [CE]. Any visible understanding of a medical product aspect starts with researchers registering experimental devices, and with manufacturers voluntarily listing physical device limitations and not performing any over-advertising. In that domain, medical product device ambassadors or brand ambassadors represent a particular gray slippery wet spot.

Medical prescription does medical prescription. Medical workplace adoption does medical workplace adoption. We could describe actual requirements in measurable dimensions [link]. What these people do, if they follow their requirements, is something they have to answer.

Physical therapy does physical therapy. Interestingly we recently saw physical therapists come out with their disgust confessions, they now need new ‘standards’ too, I guess [read statement at beginning of this post: hage].

And, user does user [link].

If I was to write a standard, it’d be a different standard for each of these groups.

Comment 2. Paper states “Research comprises a dynamic interplay between discovery and distillation into practice. So far in soft robotics, novelty presides. Little emphasis has been placed on rigorous comparisons across studies, and consequently soft robotics lacks standard benchmarks, metrics, data sets, measurement and characterization workflows, and manufacturing recipes. These challenges can be seen across scales.”

In the world of prosthetic arms, we have a stable constellation of players that all move into different directions, whereas really no one follows any orders or standards.

1. No one distils research into practice. Research is an own world where funding comes from large bodies, who do not really care about industrial / real use. E.g., [EU]. Distilling to practice would mean to actually talk to (and not down to) amputees, possibly listen to them actually correctly saying something, like, “this is broken”. Yech. Can’t have that [#vk].
2. Rigorous comparison? If you understand how no one wants to compare prosthetic arms ‘rigorously’ but how tests are rigged so the result is what you want, you can learn how to successfully wiggle in test space [reference to ansi/shap]. Consider the appearance test: a prosthesis that fulfills that would seriously sell like hot cakes, users would most likely wear only that if ever one was available – so why not make sure to stay lightyears away from that [#at]?

Comment 3. Paper states “Down at the materials level, many characterization studies use bespoke setups and procedures. Reliance on ad hoc test methods, although convenient on a study-to-study basis, has negative consequences for reproducibility. For example, measuring the resistance of stretchable electronics results in radically varied outputs, ranging from flatlined to quadratic, depending on the technique. At the structures and systems levels, issues persist. Nuances of manufacturing — casting, vacuuming and temperature treatments for soft robotic materials — vary from lab to lab, and are infrequently discussed in publications. Owing to the lack of commercially available soft robotics building blocks, labs often create and deploy soft robots entirely in-house, using custom recipes for sensors, actuators, energy sources and controllers. In other cases, the absence of standards makes it impossible to objectively compare figures of merit. For instance, terms such as soft robot ‘robustness’ are routinely mentioned in the literature, but rarely justified through quantitative metrics. When quantified, different metrics are used, including maximum strain, cycles to failure and ultimate tensile strength, precluding systematic comparison.”

This disambiguates quickly in prosthetic arm world.

Research does research. Here, the next paper is written so it follows the inside world view of academia, as remote from a user view it ever may be. They never ever, not once, apologize for any real/user world conflicting ideas [#ecr, grueber, …], so they are safe living on a different (their) planet.

User does user. We use, develop our own stuff, and we are the ones to rightfully wreck it all. We innovate way past what research or industry wants to see, and no one cares [paper]. We know how rough researchers may treat us while protecting their fragile ideas [link], so we know to not participate too much there any more. That, too. And if I want to avoid them selling my stuff back to me, I patent it myself [patent]. No one cares about our skin issues [skin] or phantom pains [phantom pains] and that is because everyone else gets to chose what they want in life (and what not).

Industry does industry. They sit on patents. They may try to dodge any costly compliance [CE] and while they may have heard of QA/QC, they may still try to work things out [e.g., link]. They benefit from built-for-obsolescence parts. They may sell as much of any old stale stuff or poor engineering parts as they can ever sell [e.g., link, and, many more, e.g. Otto Bock cable, Otto Bock wiggle hook, Otto Bock bolt/wrist, Touch Bionics / Ossur gloves, Fillauer steel cable add-on]. They may sometimes just try to be a few percent cheaper, a few percent better, than their competition and that is all there is to it.

Prosthetists do prosthetists. They buy what they have to buy because they don’t just have it lying around. Or, they just have it lying around. Then they use that. They try to comply and with that, stay within any type of limitations. They aren’t paid well. They attend industry courses. They have unhappy users, but, some may be happy though. Prosthetic arms, usually users aren’t just happy.

Circus does circus. Haha [link]. Director rules! Cripples to dance!

The real world is fragmented, shattered and utterly dystopian. We all protect core information, core knowledge. Not all methods are always explained in all detail or depth. Not all reasons are understood by everyone. Like, if one does not explain all reasons, that saves time.

Issues, yes, they definitely persist. Industry doesn’t sell well, and to show how ill-fated their attempts are they sell even less to prospective customers [e.g., Fillauer steel cable add-on]. Prosthetists also have problems, they cannot sell what users do not use. And users don’t get things they want. Researchers  created their own world, have what they call ‘healthy’ test subjects for their research ideas [healthy] and now get amputees to publicly perform in circus arenas, bringing concept of the freak show back. If one wants to do industry as industry does, they know how, they can show the way.

Comment 4. Paper states “adherence to readiness-level markers”.

Nope. Definitely not a thing. In prosthetic arm domains, readiness isn’t ever a thing, and certified readiness doesn’t even exist in the remotest dreams.

Researchers need the prototype to run exactly once at some conference demo. Then the next wild idea can be pursued. Professor Rolf Pfeifer then at the University of Zurich was said to have pursued that concept.

Industries just need the prototype to run a few times at the prosthetist before it is sold to the user. After that, everything that goes bad is the user’s fault.

Users break any device, whoever attested this to be ready.

Comment 5. What do do.

Just provide really good engineering. If I want to get somewhere as a user, I technically precisely nail my concepts and try to at least slightly overengineer my own solutions, then keep a super tight test-break-revise cycle going. That way I get gain rates that are absolutely massive. I save around 60 000 CHF insurance money over 10 years by way of pure innovation and own work in my single instance case vice versa an already cheap / affordable setup. On top, I magically generate a longevity gain of 2’700 percent when I extend my cable lifetime from 10 days to over 9 months. I was told that this damages the build-for-obsolescence plans of manufacturers, and that I undercut the prosthetists’ attempts to sell me anything, including more expensive stuff – which I may be into, buy their things, consider maybe even just, if that stuff ever it holds up. They may build whatever it is they ever want! Anyone can suggest standards or directly go into best open practice, and define a new level of quality as such. It is open to everyone else to nail concepts with precision, and, to offer me or anyone else similarly incredible gain rates, whether financial or technical. The world is their oyster as much as it is mine. Relevant here is the burning itch from hell in Charles Bukowski’s ‘excuses’.

@Article{bainessoftrob,
author = {Robert Baines and Dylan Shah and Jeremy Marvel and Jennifer Case and  Andrew Spielberg},
title = {The need for reproducible research in soft robotics},
journal = {Nature Machine Intelligence},
year = {2024},
}