Toyota Demonstrates Brain Control of Wheelchair

An anonymous reader tips us that researchers at Toyota have developed a brain-machine interface system that allows for control of a wheelchair using thought. The system processes brain thought patterns (such as the thought of moving one's left foot) and can turn them into left, right, and forward movements of the wheelchair with a delay as short as one-eighth of a second. That's a big improvement over existing systems, which can take as long as several seconds to analyze and react to the user's thoughts. "The system has an emergency stop that can be activated by the user puffing his cheeks. The BMI adjusts itself over time to the characteristics of each driver's brainwaves. If a person dedicates three hours a day to using the system, the BMI can reach 95% accuracy in a week, researchers said."

Comments

95% accuracy is pretty awesome.

[Jane Q. Public]

But hitting that wall or doorjamb the other 5% of the time really sucks.

Re:95% accuracy is pretty awesome.

[Bob9113]

I was going to say just this. 95% sounds good until you start thinking about it - but that means that in every hour of usage, the chair is going to spend three full minutes misbehaving.

OK, sure, but answer this: When you go out for an hour walk, do you ever stumble or overbalance? Sure, the wheelchair isn't perfect -- but neither are we.

Three minutes? I'd guess this thing is about as effective at understanding the brain's motive commands as an average six year old. That's pretty good.

Re:95% accuracy is pretty awesome.

[TapeCutter]

"I was going to say just this. 95% sounds good until you start thinking about it - but that means that in every hour of usage, the chair is going to spend three full minutes misbehaving. I can't find exact statistics or standards for conventional electric wheelchairs but I'd be amazed if the mean time before failure is measured in minutes rather than months or years."

Depends how you define "failure". For the type of patient that need this interface the existing interface methods would have up to a 100% failue rate simply because their disability prevents them from using it with a reasonable degree of accuracy.

Re:95% accuracy is pretty awesome.

[pinkushun]

That's 95% more than any paralyzed person can move. I'm sure this figure will improve too!

Re:95% accuracy is pretty awesome.

[nacturation]

I was going to say just this. 95% sounds good until you start thinking about it - but that means that in every hour of usage, the chair is going to spend three full minutes misbehaving.

Depends on how it fails for that 5%. If 95% of the time, it understands and executes the command perfectly, but the other 5% of the time, it doesn't understand and thus executes no command, then that's pretty good.

Re:95% accuracy is pretty awesome.

[SeaFox]

The statement was "If a person dedicates three hours a day to using the system, the BMI can reach 95% accuracy in a week", they didn't say that 95% was the highest accuracy one could obtain. After a full month of usage, you could find yourself at decimal point level inaccuracy.

Re:95% accuracy is pretty awesome.

[skaet]

Additionally, which 5% are we talking about? Does it fail every 57min for 3min reliably? Every 19min for 60sec? Every minute for 3sec? Or every 10sec for half a second?

Breaking it down like that and what do you get? A very small delay between reaction times every few seconds. Perhaps not even a noticeable delay since their optimal response time is 0.125sec

Re:95% accuracy is pretty awesome.

[BrightSpark]

The human being has a built in risk management system. I wouldn't like the 5% failure to happen say at the train platform or near a road crossing. We take special care and do a lot of subconscious checking in more dangerous situations. True we are not foolproof, but we have a lifetime of reasoning to fall back on. The wheelchair system is a bit like a child. Kids need extra help in similar conditions becasue their peripheral vision is not great, their sense of risk and reasoning is still developing. I'm sure these wheelchair owners would be making the same calls about some overide intervention at high risk times. Good idea though. I'd sure hate it if I were stuck in a wheelchair.

Re:95% accuracy is pretty awesome.

[drinkypoo]

The human being has a built in risk management system. I wouldn't like the 5% failure to happen say at the train platform or near a road crossing.

People accidentally step in front of cars and trains all the time.

Care to back up, and try again?

P.S. Slashdot's new five minutes between postings policy is retarded. I suspect that they can't afford any more servers, and the new functionality is kicking slashdot's ass, so they have to reduce the meaningful traffic. Less comments, less cached pages to generate. Only those who read and write slowest are permitted to comment at full speed. Congratulations for once again reducing the overall quality of comments with a retarded policy, Slashdot!

Wow, I have another whole minute to bitch about this, while I'm still not permitted to post my comment. I'm not adding any material to the comment, and there's still a max-number-of-comments cap, so what has been achieved? Oh wait, now I understand! Only the people who are so bent about slashdot that they will sit here and use it all day when they are supposed to be working and actually earning their pay are permitted to post a lot of comments. With a five minute cycle Slashdot has deliberately introduced a change which will reduce the number of valuable comments from people who actually have things to do. You know, the only people qualified to leave comments, since they're actually doing things?

This is not the beginning of the end, this is the continuation of the end. The beginning of the end was requiring people to take one minute to post a two line comment. I think I see the drain getting closer, but it's hard to tell because it keeps going around and around...

Thought / Action Barrier

[wasabu]

Uhmm.. what happens if you can't stop thinking about moving?

Re:Thought / Action Barrier

[interkin3tic]

There's a difference for most of us between thinking ABOUT moving our foot and thinking TO move our foot.

Wheeee!

[powerslave12r]

I can't wait to hook it up to a Wii and play some racing games.

For those who RTFA: Though-Controlled Robot!!

[Anonymous Coward]

If you actually went to the article and watched the video, there is a side-note at the end that Honda developed a similar system for controlling a robot.

They demonstrate using ASIMO.

HOW IS THIS NOT THE MAIN STORY!

Re:95% accuracy probably not good enough

[Korin43]

That seems to be why there's an emergency stop built in..

Gah... brains are meant to be good at learning

[gringer]

Okay, I'll say this again, because it doesn't seem to have filtered through to the general population yet. Until the singularity, the human brain will be able to learn more easily than a computer. Please stop trying to teach computers the thought patterns for specific movements, and just provide a neural interface for the brain to work with. The brain will be able to figure out what signals it needs to fire to get the wheelchair moving (or whatever) soon enough.

FWIW, my own idea of how to do this would be to put a few small electrodes into a person's lower arm, far away from the brain (and have a sensitive meter to detect nerve firings). Once the brain figures out what nerves are important for this interface, you then use that interface to deliver signals to operate other equipment.

Re:That's nearly perfect.

[JanneM]

"Oh, good, I stand a relatively good chance of being able to stop myself before rolling into traffic."

Which is why the brake is controlled with the breathing tube.