Voice Controlled Wheelchair

[mgwmgw]

Our current plan is to use a voice recognition similar to how Amazon's Alexa can recognize a voice command and respond with feedback. Our goal would be for a user to be able to call the wheelchair to come to them if it is not beside them. Once in the wheelchair, the user will be able to give a command to take them to a destination inside their house.

https://www.todaysmedicaldevelopments.com/article/autonomous-design-manufacturing-wheelchair-kd-smart-chair-111618/

Can we anticipate the mayhem from this? How about if a dog bark is understood as a command?

[Burgerman]

Voice control on my PC is rubbish. Same as that fitted to cars! I hope this is an awful lot more accurate. Or we will see them under buses, in rivers or through plate glass windows! Even that simple command and stuff you get on phone merry go rounds never works.

[Fedor]

Right direction for research. But voice control is bad for that. Human speech is very long for such purposes. Perhaps it will work well to control the lifts or lights, but no more. The article also talks about LIDARs and 3D cameras, it sounds much more interesting and realistic.

I do not understand why I should always keep the joystick on the 2 kilometer track, even when my nose is itching or i want to drink. So some autopilot would be cool.

[Burgerman]

Used to have 1 voice activated thing that worked! My keys finder. A cheap plastic thing that made the lost van or house keys whistle back. You whistle, and they went off. Easy to find keys.

Even that had a downside. The stupid dog never came when you called. Or whistled.

But it took about 2 weeks before he figured out that when I whistled and the keys went off, that I was leaving the house! So the dog appeared at around 40mph crying and whining barking and shaking with excitement. It meant he might be going out!

So that went in the bin.

[Burgerman]

Like these. Mine had no light. Not exactly the same as driving a chair...


youtu.be/RAie70LBGRw

[ex-Gooserider]

Basically old news - how many student groups that want to 'help' us have shown up here with essentially the same idea???

While something of the sort might be handy, sort of the way 'cruise control' is in my van, I don't see it as all that useful for most of us, at least those who aren't at the Rachi or Steven Hawking level of disability - where almost anything increases mobility abilities... A self driving chair might be of some help for that group of users, in that they no longer need an attendant to drive for them, but I wonder how much help it is if the person is still going to need an attendant to do other things for them at either end of their travels?

It seems the prevailing mentality of these researchers are that we are helpless objects that need to be carried around like cargo.... I am a PERSON, not a potato, thank you very much, and can drive myself....

I don't need a 'better' interface - I need a chair that can do better things with the interface I have!

I didn't need a self driving chair a few nights back when I got totally stuck in the 6" of 'global warming' on our driveway and access ramp - I needed a chair that could drive over / through it when I wiggled the joystick....

I can imagine that there might be a better interface than a joystick (I don't know what it might be) as a way of transmitting my desired movements to the chair - but *I* want to be the one doing the driving.... Getting turned into a blob to be driven is NOT the solution....

As my friend Erik says in his pitch for the hoverboard project - I want to be the PILOT, not the passenger!

ex-Gooserider

[Burgerman]

As my friend Erik says in his pitch for the hoverboard project - I want to be the PILOT, not the passenger!

Changing the subject a lot, most airliners have "pilots" with very low piloting skills, and are in fact traveling much as cargo themselves. A modern airline pilot can often get just 5 mins actual flying time doing the same repeated stuff, in a 10 hour flight. So in a 1000 hours of time on a aircraft may be just a couple of actual hours. As such theres a lot of frankly crap pilots on airliners that are basically bus drivers that are screwed if anything goes wrong. Made worse if anything mechanical happens as most dont understand basic physics, or mechanics, or how their own aircraft works. How do I know? I often talk to the KLM pilots in my local hotel bar... Its frankly amazing how little they know, and I am just an unskilled drinker!

[foghornleghorn]

Voice control is going to cause accidents.

I quite like the idea of a wheelchair that could be instructed to autopilot take me home when I'm knackered though.

[ex-Gooserider]

Quite true - a few years back there was a big Air France crash... The post analysis said the initial cause was a sensor getting iced over, causing the autopilot to put the plane into an excessively steep climb.... But put the crash 100% as pilot error due to failure to properly recognize the problem and properly respond to it....

The engineer writing the paper I read on it, said that the real issue was that the pilots had so little experience of actually FLYING the plane that they didn't have the skills required to recognize and respond to the message they were getting from the instruments the way that a small plane pilot who is doing all the flying manually would have... The autopilot had found a 'corner case' where it's programmed behavior made it do the wrong thing, and instead of over-riding it and correcting the situation, the pilots spent considerable time on discussing why their instruments were acting oddly... By the time the senior pilot actually did start the right action, it was to late for it to save the day...

The paper suggested that this may become a problem as we get more 'self driving' cars - that while the car may do a fine job in 'normal' conditions, it will lead to drivers not having any 'situation awareness' and being unable to deal with the problem when the fecal material impacts the rotary ventilator... My only problem is wondering if it will make that much difference given the typical driver's (lack of) skill level...

ex-Gooserider

As my friend Erik says in his pitch for the hoverboard project - I want to be the PILOT, not the passenger!

Changing the subject a lot, most airliners have "pilots" with very low piloting skills, and are in fact traveling much as cargo themselves. A modern airline pilot can often get just 5 mins actual flying time doing the same repeated stuff, in a 10 hour flight. So in a 1000 hours of time on a aircraft may be just a couple of actual hours. As such theres a lot of frankly crap pilots on airliners that are basically bus drivers that are screwed if anything goes wrong. Made worse if anything mechanical happens as most dont understand basic physics, or mechanics, or how their own aircraft works. How do I know? I often talk to the KLM pilots in my local hotel bar... Its frankly amazing how little they know, and I am just an unskilled drinker!

[Burgerman]

I remember that crash. They stalled the plane, all the way down to the sea. The pilots figured out they were stalled seconds from impact. At low level they applied power and forward stick... No basic flying skills.

Never occurred that they were flying with high angle of attack. With part throttle. Which isn't possible. For about 14 minutes. And were dropping!

[ex-Gooserider]

Exactly - the sort of situation that a manual pilot would have not gotten into, but because the autopilot was driving, they didn't realize what was happening.... The autopilot screwed up because of the sensor problem that gave it bad info....

Way back in my college days, the school had a flight school and a policy that non-flight major students could get a free ride in the back seat on a training flight - (note the rules said it had to be 'instructor training' - not novice training)

I took them up on it, and part of the flight was what the instructor called 'unusual attitudes' I learned about this when I was asked if I got airsick! I was then told to let them know if I started having problems as they didn't want to have to clean the plane...

The training was that the 'student' would have to wear a hood that blocked outside vision, and look at his lap so he couldn't see the instruments... The instructor would then put the plane through a bunch of turns and spins to the limit of what aerobatics it could do, then tell the student to take over... The student would have to figure out what the plane was doing, correct it and return to 'straight & level' flight on the designated course just by looking at the instruments.... Definitely a fun ride....

The Air France pilots hadn't practiced this enough (especially on that plane) to recognize it fast enough...

In the car driver case - supposedly the typical accident scenario has less than 3 seconds between the initial 'Oh Crap' and 'Bang'... If the self driving control messes up, how many drivers would even be able to look up and figure out what is wrong let alone respond to it??? When I am driving, I at least know what the car is doing (hopefully...)

ex-Gooserider

[Burgerman]

Years ago I used to fly or at least learn to fly sailplanes. Gliders. With both aerotow and winch launch. And in everything from the old auster tow plane (always on an odd number of cylinders) and high performance modern sailplanes to 1935 built T21 and T31 open cockpit machines. I was taught to recognize airspeed and angle of attack and the likelihood of stall by the change in sound as the airflow angles and airspeed changed. And those things even apply to an airliner. And to back that up with sluggish feel on controls, adverse yaw increase on aileron, and more elevator being required to keep constant airspeed (sound) etc. And was taught to use these sounds and to listen and observe everything. And the control "feel", and to be aware of pre stall buffet which you feel when on the edge of a stall, even at high speed if angle of attack is too high. And another half a dozen speed clues.

We were allowed to purposely stall the things at altitude to learn, over and over by pulling too many G's in turns, by flying too slow, applying too much aileron, etc. By these means we learned how a stall happened, So we could drop the nose or not apply too much control, on the edge. So learned how to recognize the situations and control movements, and so on that caused a stall and what kind of stall. In the end you can throw the thing about on a wing tip pulling enough G to initiate pre stall buffet as you are coming in low, to NOT actually stall it and die, with confidence.

This stuff is all backed up with thousands of hours flying models, of every type. And these respond the same way. So I would not get into that situation myself. And what do I know, I am a plumber! You would EXPECT airline pilots to understand and have a natural feel for this stuff. But having flown with a few, and spoke to many at the bar in my local hotel, I KNOW that they dont. They are trained like bus drivers. They are unaware of many of the things that a light aircraft enthusiast takes for granted. And unless they are ex military or have a flying hobby, even with models, they do not understand much. Esp how their planes and engines actually work... Things like the effect of CG position. They really dont get what it does to the feel. Or the stability. Or lack of.

[MichaelB]

Popular Mechanics published the Air France crash report https://www.popularmechanics.com/flight ... 7-6611877/ it resulted in different training for pilots. I was on a flight to Hong Kong and sat next to someone handling Air France PR and they said that it was a difficult situation as the pilots didn't know what they were doing.

Boeing going through a similar thing with 737 max/Lion Air crash, faulty sensor pilot training error https://www.youtube.com/watch?v=sgCK5wbct6I

[Burgerman]

Seems to me that you cant "train" this sort of thing. You either naturally get it, have an enthusiasm, and flying interests and hobbies or you will always be a bus driver. In fact they dont do as much as a bus driver. There are pilots that have only done 2 hours actual hands on flying in a year of flying across continents.

No I think that they need better assessment of natural ability and flying capability. Raise the entrance bar. They need the sort of guy that does aerobatics in his own plane at weekends. The sort that build their own planes. Even hobbyists. I can design, scratch build a turbojet powered 300mph hobby jet that fly well straight off the building board. I know how they fly, what causes control flutter, where the CG must be, what shape/weight it must be to perform well, Where it needs strength, and be stable at high speeds yet have enough control and glide capability to be easy to land with a dead engine. I built pylon racers, canards, deltas, pushers, multi engine stuff, and everything in between. Out of my head. Since I was a kid. Before there was easy to get RC systems. As such I found gliding, full size simple in my teens, and wasn't dangerous. I understand the physics, the aerodynamics etc. More so than the instructors. So I found it easy and quite natural.

When choosing airline pilots,they need to "train" people like this above. I haven't a clue about rules, radios, automated computer systems, and systems designed to "help" the pilot. For E.g. In one air accident I was reading about, a lady pilot clipped a wing on the floor in a modern airliner and almost crashed. As it was there was extensive damage to a wing and she ran off the runway. Why? Because the stupid designer had REDUCED aileron authority (and aileron control sensitivity) by some 50 percent electronically as soon as it sensed the aircraft wheel touch the runway to prevent "over control" after touchdown. Well it was windy, and a crosswind. She NEEDED that aileron control! And it wasn't there. There is hundreds of such stupid systems on modern airliners that should not be there. The fix? To inform the pilots of this feature. So they dont make her mistake. I kid you not. My own model planes do the OPPOSITE. At low speeds you need MORE control authority, as the plane is slower. And you might need it on landing due to turbulence etc.

[Burgerman]

: Regarding this : 737 max/Lion Air crash

In any light aircraft, or a glider, or the hobby jets I build, we purposely have the CG in the correct position, to give a marginal trim change with speed. Because having CG in front of CP also means natural pitch stability. Its secondary very useful purpose is control feel. As the aircraft slows it naturally drops its nose. So that you need a little back stick pressure as speed decays. This ALONE is all you need to get a good idea of actual speed. But those clever airliner designers have added automated electronic trim control so that this natural feedback system is missing. As you slow even a nose heavy airliner, there is zero trim change. So you dont feel the loss of airspeed. That is dangerous!

They also try to load up the plane so that this natural stability is gone, by allowing a very rearward CG so that CG and CP are in the same area. That saves a few percent fuel. But loses out on the natural aircraft stability. That pitches it nose down as speed falls. That would have prevented a long stall into the ocean on the air France flight just by itself. It would have entered a nose down position if the wing stalled on its own and would have naturally corrected the stall. So for the sake of a tiny fuel saving they removed the natural stability and rely on instruments. The difference is small but significant.

[MichaelB]

The problem is that statistics will show fewer accidents on modern planes equipped with safety features that reduce pilot input. Fewer accidents = they are doing the right thing. Of course you don't have data to know what would happen if they removed all those features. It'll be the same with self driving cars when they become mainstream.

[ex-Gooserider]

I don't have the background to comment on the aircraft design side, though what BM says makes sense to me... But IMHO more important, at least in a plane, is to require the pilots to have enough experience driving on 'manual' to be able to take over when the autopilot does the wrong thing...

I'm not sure it is possible to do this in a self-driving car, just because there is so little time between problem and crash that there really isn't time for the driver to intervene effectively...

So what I think will end up as the case is that MOST of the time, the self driving car will be better / safer than a manual driver, but it will make messier accidents in the 'corner cases' that it doesn't handle correctly... Also if the car does allow 'manual mode' driving in unusual cases there will also be more issues due to the driver not having had as much practice...

ex-Gooserider