A model trained in a virtual environment does remarkably well in the real world.
See full article...
See full article...
AI helps exoskeleton adjust to different users, handle stairs
- Thread starter JournalBot
- Start date
No photos? I mean, I love the Aliens reference, but it's hard to understand the article without at least one photo of the exoskeletons they're creating..
Upvote
67
(68
/
-1)
The Best Bionic Leg Yet
A surgical procedure and muscle-sensing electrodes allow neural control of a prosthetic limb
spectrum.ieee.org
Last edited:
Upvote
7
(9
/
-2)
Exoskeletons for people with disabilities? Is Su's team not only creating power armor, but also dreadnoughts?
Upvote
16
(17
/
-1)
Is that actually what this article is about? Neural interfaces for prosthetics are fascinating and promising, but don't really sound at all like AI-trained exoskeletons.A video and pics over at ieee
The Best Bionic Leg Yet
A surgical procedure and muscle-sensing electrodes allow neural control of a prosthetic limb
Upvote
20
(20
/
0)
It seems tangentially involved, but I would allow it. Interest in robotic exoskeletons would easily lend itself to robotic prosthetics and, perhaps, artificial organs.
Upvote
-12
(1
/
-13)
Ahh shit, you are correct, I misread the Nature articles and thought they were related.
Upvote
6
(7
/
-1)
AI meant to help humans do their job, not replace humans at their job, is the type of AI we should be focusing on. Glad to see a project on the good path.
Upvote
18
(20
/
-2)
Upvote
1
(2
/
-1)
I had a teacher who traveled and played a surrealist game of Surautomatism on nice evenings. He would leave his rooms and turn left or right without thinking. At the next juncture, he would repeat adding options as the juncture legally allowed. (He obeyed traffic laws and did not permit himself jaywalking).
He would do this until he met someone who ended the game.
Will there be an exoskeleton for him?
Upvote
-11
(2
/
-13)
Ripley : Hey, I feel like kind of a fifth wheel around here. Is there anything I can do?
Sergeant Apone : I don't know, is there anything you can do?
Ripley : ...Well, I can drive that loader. I have a Class-2 rating.
Sergeant Apone : Be my guest.
[Ripley mounts the second loader, activates it, then casually drives it to pick up a cargo container]
Ripley : [nonchalantly] Where do you want it?
[Hicks grins, and Apone booms with laughter]
Sergeant Apone : Bay Twelve, please.
edit: credit IMDB.
Sergeant Apone : I don't know, is there anything you can do?
Ripley : ...Well, I can drive that loader. I have a Class-2 rating.
Sergeant Apone : Be my guest.
[Ripley mounts the second loader, activates it, then casually drives it to pick up a cargo container]
Ripley : [nonchalantly] Where do you want it?
[Hicks grins, and Apone booms with laughter]
Sergeant Apone : Bay Twelve, please.
edit: credit IMDB.
Upvote
28
(31
/
-3)
With all the misuse of AI and ML these days, it is kind of refreshing to see an article and go, hey, that's a great application!
Upvote
15
(18
/
-3)
I kinda wish they elaborated more on the applications for assistive devices for those with disabilities, but I suspect that will be a series of future papers. Both the quote in this article and the brief mention in the paper makes it sound very much like an open question, despite their confidence:
Like would you even want to directly model the gait of someone with knee arthritis? Their gait is already modified to reduce pain or stress, and might be deeply sub-optimal for an assistive system. It might be more ideal to restore the natural gait that existed before arthritis. Something like cerebral palsy would be even trickier since the system would have to guess at and assist the intended motion, even given deficiencies in neuromuscular control. I have to imagine that the approach would need to be tailored for every disabled individual; maybe this could still help even if the process is no longer completely "experiment free"? A fitting session on a treadmill while assistance is dialed in by a physical therapist wouldn't be unreasonable, and isn't so different from fittings for other assistive devices like hearing aids.
Still amazing work and very promising!
Like would you even want to directly model the gait of someone with knee arthritis? Their gait is already modified to reduce pain or stress, and might be deeply sub-optimal for an assistive system. It might be more ideal to restore the natural gait that existed before arthritis. Something like cerebral palsy would be even trickier since the system would have to guess at and assist the intended motion, even given deficiencies in neuromuscular control. I have to imagine that the approach would need to be tailored for every disabled individual; maybe this could still help even if the process is no longer completely "experiment free"? A fitting session on a treadmill while assistance is dialed in by a physical therapist wouldn't be unreasonable, and isn't so different from fittings for other assistive devices like hearing aids.
Still amazing work and very promising!
Upvote
7
(7
/
0)
Did you hear about Helen? Yeah, that’s right. They put her in one of those new suits they leased, the ones with the big claw things and the winches. No, no, nothing like that. You’d think so though, right? But she was wearing it all day and the thing was working fine and then it just folded itself up while she was walking back to the charging station when her shift ended. Yeah, I overheard the techs say that the suit must’ve seen one of the signs for the loading dock and assumed it was going into a shipping container so it put itself into travel configuration. Oh, no way. No way. Like a balloon full of jelly, man. Closed casket for sure. They said there’s probably no way to tell what it was in the training data that let it bypass the safeties. Hell of a thing.
Upvote
28
(34
/
-6)
This is amazing! I'm excited to see what innovation is spurred by this. AI enabled human interfaces will be a much safer and more secure evolution than neural links for physically able people.
Upvote
0
(1
/
-1)
Nah, gramps lost his little finger in the war to annex Canada, a lucky shot hit the joint on his T-45.
War. War never changes.
(No idea why he's watering the flowers with his own blood though.)
Upvote
13
(15
/
-2)
So they’re still dealing with at least coarse state machines, right? Walking, running, going to stairs, vaulting an engine-killed APC to surprise light infantry on other side, etc etc.
Upvote
11
(12
/
-1)
The article doesn't explain why those systems need to infer the intentions or why they need any kind of tuning. If the system provides adequate feedback and has enough degrees of freedom for the human operator to use intuitive stabilization, then wouldn't the human operator be able to just wear it (possibly after short learning on the human side)? Is it so prone to PIO that its inherently unstable?
Upvote
1
(3
/
-2)
Just to be clear, "helping humans do their job" does result in either less jobs, or more productivity. And, most of current AIs cannot really "replace" anyone, but does help them in their jobs, which can mean less jobs.
Upvote
-5
(1
/
-6)
Upvote
1
(1
/
0)
Those of us who live in the state will not recognize such an institution. North Carolina State University is the proper name for the institution. He is also associated with the University of North Carolina at Chapel hill through their join biomedical engineering program. Similar distinction between University of Michigan and Michigan State, University of Virginia and Virginia Polytechnic Institute and State University, Texas ... . The names are close, but the distinction can be important.
"Dr. Hao Su, PhD, is an Associate Professor in the Department of Mechanical and Aerospace Engineering at the North Carolina State University. He is also a faculty at the Joint UNC/NCSU Biomedical Engineering Department. "
Upvote
13
(13
/
0)
I don't think that's blood... There's other fluids that could work well and might be easier to get to.
But still, many iteration digital modelling seems to work well in other fields, so why not exoskeleton/prosthetics?
Upvote
1
(1
/
0)
For the elderly. Imagine being able to rent an exoskeleton in Rome, Xi'an or Angkor to visit the huge ancient ruins there without having to use a wheelchair to get around.
Upvote
3
(3
/
0)
This article is so confusing. It isn't until the last paragraph that it says it's about MEDICAL exoskeletons to help mobility impaired people. And I only noticed that after seeing another comment and reading the article again.
This matters because until getting to the point that states this is a medical exoskeleton, this looked like an over-engineered solution for something that an analog system would be more robust and give the user a faster response time.
This matters because until getting to the point that states this is a medical exoskeleton, this looked like an over-engineered solution for something that an analog system would be more robust and give the user a faster response time.
Upvote
0
(0
/
0)
This seems like a better use of AI than a lot of what i’m seeing.
The robots were supposed to do the menial chores like dishes and mowing the lawn so I could spend time on art and music.. not putting artists and musicians out of work leaving them to seek employment as dishwashers and custodial staff.
The robots were supposed to do the menial chores like dishes and mowing the lawn so I could spend time on art and music.. not putting artists and musicians out of work leaving them to seek employment as dishwashers and custodial staff.
Upvote
1
(1
/
0)
