# What (if any) are the engineering limitations for producing “simulation” exoskeletons

To clarify the title since it cannot be said within the word limit:

## A "simulation" exoskeleton, is an exoskeleton that doesn't exist to support a disabled user's movement or to help someone carry more. It is an exoskeleton that one wears in order to be better immersed into a virtual environment, either for recreation or training.

To explain further, the increased immersion is achieved using the exoskeleton for mechanical feedback, and for full body input, by matching the "rag doll" avatar's position/form to the exoskeleton you can both apply more diverse control to and experience the virtual environment in a far more immersed way. Allowing running, wrestling, climbing, carrying, etc.

That's the theory at least, and products like Dexmo achieve it on a small scale.

And there was a product in progress called Axon VR which would have been a full body experience, I can't find what happened but it seems they've decreased scope and are also doing gloves. The shtick of these gloves compared to Dexmo is that there is better tactile feedback, such as heat/texture/etc. Axon VR even has an ex-CEO of twitter invest in it so obviously the concept isn't too out there.

Exoskeletons like these that can be mounted and so don't have to lift your or their own weight, nor have to worry about batteries since then can run off the mains, should be significantly easier to make economically viably right?

So I want to know

A) Whether something is feasibly possible with a 100,000 dollar hardware budget per item at all

B) If not what is limiting it?

C) If it is, then how much would it cost (ball park numbers), and a rough cost breakdown in terms of hardware per complete item.

I understand that software for it is hard as hell, and ensuring safety within legal limits is difficult, or perhaps there isn't a big enough market that even if they were $25,000 to buy each that they wouldn't sell. By answering the above questions I hope to narrow down on the "bigger picture" of it. From what I read online I could find upper limit of power outputs of humans reached around 2000kw (~2.5hp) or for a short period for a professional athlete, in this case cyclists. A 5hp pump can be bought for under \$1k off the first site I found, even if you had to go as high as 10hp to not only stop but also "reverse" any actions of a "sprinting" human, which probably wouldn't be safe anyway and wouldn't be a feature, as well as being extremely rare, that's sub \$2k. After that you have material for a frame, ~40 cylinders/hydraulic motors (total), hoses, sensors, and (most expensive as far as I know) ~40 servo valves, and I'm sure some other things. (~40 estimated as 3 for each shoulder and hip joint (+12), 2 for each upper and lower arm and leg (+8), a few for each wrist/ankle, and 5 for each hand (+10)). Now I admit servo values can be PRICEY, but in bulk and many of the 40 not under much stress like the ones for fingers I figure the price should stay WAY under \$250 a valve + cylinder, which even then becomes \$10000 which for what you're getting isn't too high in my opinion. And that's an all out all motion one, I imagine most people could have lots of fun with just hands, feet, and waist "simulation". I mean there are already exoskeletons on the market that can walk people who can't walk! On battery, against the force of gravity, within a reasonable price range. I'm honestly mostly wondering if I'm crazy more than anything, wondering why we don't have simulation exoskeletons already, maybe I am just vastly over estimating their market value. ## 1 Answer The answer is fairly simple: • for data input, video recognition is far less cumbersome than an exoskeleton reading the user moves. • For output - force feedback - our senses are way too complex; the technologies currently provided are all snake oil and money grabs because they would be extremely lackluster had they came to fruition in current form. The problem is that these devices provide force feedback at bone/joint level. This is not how humans perceive world - we have a huge number of touch receptors in our skin, and they are the primary means of giving us tactile feedback. When you push a wall with your hand, you expect to feel pressure on your palm and bottoms of fingers, not a pull on their topside. You feel the palm resisting as your wrist, elbow and shoulder strain against the resistance, not pressure of straps on your arm, forearm, and a pull on top of your hand, with again, no pressure on palm! That's a completely different sensation, not even remotely similar to the actual experience. You'd need an exoskeleton that encloses your entire body tightly, and is able to apply resistance at any point while not obstructing any other - you couldn't use the same 'bones' of the exoskeleton to provide sensation to forearm and simultaneously support the bones of your palm, because if an obstruction blocks the palm, the skeleton must transfer the force all the way from the palm to where the equal and opposing force is applied - at foot soles(!) without affecting any other body part in the process! If it makes the elbow joint rigid, you're surely feel it in your arm and forearm, not just in your palm. The price and complexity of even a simple, flawed exoskeleton is huge, and the result so lackluster and disappointing in comparison, that not a whole lot of people are willing to invest into it - and most of these who do don't realize the flaws. • I think I'd have to experience it to agree you are right about it feeling too wrong to be "worth it" but I certainly can' deny that it is a likely reason. Perhaps just simulating flat surfaces for hands and feet combined with Dexmo/HapticX style gloves would be satisfactory. – Shefeto Jun 7 '18 at 11:41 • @Shefeto: Then these will still remain flat surfaces. You won't be able to, say, climb a rough wall using cracks in it. You won't be able to grab a corner or edge of a table. Better don't try sitting on it. We're much better off just skipping exoskeletons and trying to beeline to the neural joints where you could bypass the receptors spread all over the body and just input the sensation into the spinal column. We're nowhere near but it would be about the only way to get the "real feeling." – SF. Jun 7 '18 at 11:59 • I just think it'd still be better than no feed back which is the current option. And it doesn't have to be a flat surface, you could have a couple things like a hilt/handle and immediately you can simulate most sports with a bat/racket, add a gun handle you can simulate a lot more things and so on, they could all plug into a common motion device so you'd ultimately only have to buy new trinkets or maybe even 3d print your own. Maybe you can't climb a heavily textured wall super realistically...sucks but not idea crushing I don't think. – Shefeto Jun 7 '18 at 12:05 • @Shefeto: "I just think it'd still be better than no feed back which is the current option." - better, yes - but you should always think in terms of costs vs benefits. Having a lunch and a glass of water is better than just having a lunch without the drink, right? Then consider, what if the lunch is \$10 and the glass of water is \\$500? Is still the lunch+water better? (and no, you're not in a desert dying of thirst or anything like that.) – SF. Jun 7 '18 at 12:26
• But in this analogy I would be dying of thirst with how desperate I am for improved immersion ;) – Shefeto Jun 7 '18 at 12:52