Hacker News Reader

Comment by amiga | Hacker News Reader

I'm a guy who has disassembled and reverse engineered a standard Jazzy power chair, and what I noticed was the attention to detail regarding failures. The chair is thoroughly designed to shut down at the slightest bit of trouble. There's some redundancy in things like the controller, where it used redundant hall effect sensors that were identical to the others, but ran in an inverted power profile, to detect any weirdness in the sensor outputs.

I ended up adding a long range remote control to it. A remote control power chair is fun to drive around. People do get a little concerned when they see a chair rolling around without a driver

View on HN

My mum recently had a curbside crash while she was riding an e-bike. This resulted in her breaking bones in both her hands, which resulted in a surgery in her left hand and various problems (tcl fracture related) with her right hand.

This makes me actually appreciate reliability in e-vehicles motor cutoffs etc. I keep thinking if this could have been avoided with a better quality e-bike or if actually it would be even worst with a cheaper one.

Which makes one think, how often a wheelchair with cheap e-scooter parts would crash people into staris, cars etc

I know public use devices have their own problems with reliability, but I did almost cause a traffic accident a couple times over the years. Every time, the scooter's accelerator lever got "sticky" due to repetitive (mis)use, and would sometimes not go all the way to 0 when released. Stuck at ~10%, the scooter would brake normally and remain at halt under my weight, but the moment I stepped off it, it would suddenly launch itself at the cross traffic.

It's these little things that get you. The scooters all have some kind of debounce logic, disabling the accelerator until you're moving sufficiently fast - but the logic doesn't kick in when you stop without releasing the lever. A little bit of redundancy would've helped here.

A friend has an e-unicycle (I think the category devices has some other name as well..) and he wanted to try out how it behaves in a track.

He sort of knew, but didn't expect it, that when the roll of the device exceeds a certain threshold, the device will shutdown. Even if you're on a curve going with some speed. Broke his wrist. Since then he's also wearing wrist protectors that keep the hand straight.

Actually it was a bit unexpected that it would have known to do that; it must have used its complete IMU data to even know it was rolled, as plain accelerometer would have been pointing "down" as usual.

> Actually it was a bit unexpected that it would have known to do that; it must have used its complete IMU data to even know it was rolled, as plain accelerometer would have been pointing "down" as usual.

That actually feels like overengineering based on well-intentioned, but wrong specs. You probably want to just use sideways acceleration for "falling over" detection, instead of roll.

I'm an embedded software engineer with past experience developing robotics and motor control drivers.

Those e-unicycles terrify me. No way I'd trust my life to one. Once you're at speed, every failure mode results in instant passenger ejection. I see people flying through traffic on those things - they're just one sensor glitch or integer overflow away from serious injury.

The safety with ebikes does vary a bit although I'm not sure it's down to price. My one is quite a cheap one but has quite a lot of safety features - will only go if you pedal it, motor cut if you touch the breaks, 14 mph speed limiter etc. But I guess you can come off any two wheeled vehicle.

> People do get a little concerned when they see a chair rolling around without a driver

Add a hat and a scarf on a wire and you've got a Halloween prop.

Reply on news.ycombinator.com