Regarding quadcopters, this video by "Verity Studio" https://www.youtube.com/watch?v=3h1Kh152ygU claims that they have developed a technology to enable drones to continue to fly a little, and then land safely, with just three motors and the fourth disabled.
But to the best of my knowledge this hasn't been implemented anywhere, and today's drones simply fall out of the sky if one motor fails or its propellers are destroyed.
Anyone know why that is? Is the video simply fake? (Comments on the video are disabled, which is not a good sign...)
Even if the video is fake, this seems like an important area of research for drone safety; what's the state of the art?
> But to the best of my knowledge this hasn't been implemented anywhere, and today's drones simply fall out of the sky if one motor fails or its propellers are destroyed.
I'm a former Skydio[1] employee and our motion-planning & state-estimation teams implemented this on all our drones years ago. It wasn't completely fool-proof, but in many non-high-speed scenarios a loss of a motor was not catastrophic and the drone could 'emergency land' within a few meters of where it was flying. I wish I had proof for you but unfortunately no longer have access to any internal videos :)
I remember a paper/ video from ETH Zurich demonstrating this sometime around 2015. There's nothing particularly difficult about it if your motors and controllers are sufficiently overspecced and you're willing to let the aircraft start spinning (you sacrifice yaw control).
I don't think so. At least there is nothing impossible about what they are claiming to have achieved.
> Anyone know why that is?
I think the problem is that it wouldn't gain you a new ability. If you are working on something where resilience against failures is important you are much better off with an octo- or hexacopter arrangement.
With that algorithm if a failure occurs suddenly you are in a brand new control regime where the pilot is probably not as proficient. How will they know which way to push the stick when the drone is spinning like crazy? So you probably want a fully autonomous controlled drone (like the ones in the theatre production they are showing).
> today's drones simply fall out of the sky if one motor fails or its propellers are destroyed
> it wouldn't gain you a new ability (...) suddenly you are in a brand new control regime where the pilot is probably not as proficient
It would be an incredibly powerful marketing message.
The pilot doesn't need to be able to control the drone; the drone would simply land where it is, which, depending on the location, could be not ideal, but in any case much better than falling like a brick...
But to the best of my knowledge this hasn't been implemented anywhere, and today's drones simply fall out of the sky if one motor fails or its propellers are destroyed.
Anyone know why that is? Is the video simply fake? (Comments on the video are disabled, which is not a good sign...)
Even if the video is fake, this seems like an important area of research for drone safety; what's the state of the art?