The emergency fault problems you've described are just the tip of the iceberg for "one motor per wheel" designs in real life. At low speeds things generally behave... but at anything approaching highway speeds, this will require real-time active mixing of the motor ESCs similar to that used in a flight controller.
This is something I have first-hand experience with trying to build large-scale AWD stadium trucks. ...
I have no experience with per wheel direct drive motor designs in vehicles. And I can't help with mass placement optimizing and so on. Anyway, I do know what modern industrial motor controllers can do, and that's way above what your flight controller in a drone can do, or in any (larger or smaller) model does. Electronic gearboxes are just one of the rather well-known and simple applications. So real time emulating the behaviour of a 2WD differential, or multiple differential (and locked) 4WD is nowadays nothing too exciting for these controllers. The system should also be able to actively steer the "emulated" mechanical differentials in more ways than the mechanical ones, intended as driver assistance. It's perfectly possible, but a big heap of work for the engineers to correctly design and validate this so the vehicle doesn't behave unexpectedly.
This is exactly what I'm getting at. It requires
active management to work at highway speeds. I said
similar, not the same; as in working in a similar fashion, with accelerometer/gyro closed-loop feedback, etc. So now the big problem... one you are
not going to be able to get around... is how are you going to get the vehicle to a safe halt if there is a catastrophic failure in the electrical system
at highway speed. Something that takes out the CPU which controls the mixing. Now you are trying to bring to a halt a vehicle where each wheel is literally pulling the vehicle in a different direction due to torque reaction.
EDIT: And here's another, much more likely scenario: One of your OMPW motors overheats, maybe burns a winding and dead-shorts at highway speed. This has a high likelihood of taking out at least part of the electrical system, if not all of it. Now you have no active control over the remaining wheels, and you have one wheel trying to pull the whole vehicle off the road due to the EM braking we've been discussing so hotly.
Same scenario with one motor per axle, and the diff handles all the torque balancing inherently due to its nature, and the motor drag is just additional braking bringing the vehicle to a stop.Even with whatever "secondary system" you are trying to use as a emergency backup, you cannot guarantee any measure of control even with "backup on a backup" redundancy. Having a completely separate system based on a completely different technology which can bring the vehicle to a halt in the event of no power/no processor really is
the only reasonably safe way.Don't get me wrong; I'm all for making regen braking take over as much as possible from those mechanical systems... a finger in the air WAG says taking over as much as 90% of braking is probably possible, and that's a effing
lot.But trying to take it all over and eliminate mechanical braking, or replacing what is already a very simple, very efficient differential system and increasing electronic complexity 1000-fold is not the right way to do it, IMHO. This whole concept smacks of
"just slap a µ-processor in it" thinking, which is something we very often rail against in here.
And add to it the fact we don't even have appropriate motor technology for this yet... come on. I'm all for conjectural conversation, but I think with the brainpower in here, we should be able to come up with something that can actually be made real in the decade or two I have left, which is where we
need it if we are to avert the global doom we keep averting our gaze from.
As an aside... if you weren't aware, modern high-performance quadcopter FCs are actually pretty sophisticated now, and it is pretty amazing the feature-set they can cram into a $40 PCB.
We're talking fast enough, granular enough control of the ESCs which on acro/racing FCs, even ground-effect propwash can be tuned such that it is almost completely eliminated.
That sophistication is actually part of the problem, and I suspect is most of the real reason the FAA and other similar bodies are trying to take over every aspect of model aviation right now;
you can buy a FC and GPS receiver off of Amazon for less than $100 with the capability to pilot a vehicle (multirotor, rocket or fixed-wing... even land/water craft in principle) to "within a dozen meters" accuracy from one continent to another. All you need is to fix the firmware so that GPS-based "no-fly zones" are ignored and "deliberate misinformation" offsets in the GPS network are corrected for. In principle, this has been true for over a decade; the first 32-bit GPS-equipped FCs were released that long ago.
Let that sink in for a second. Honestly, it surprises me that so far, we mostly only have drug cartels using that tech to get cocaine across the border. Obviously, for passenger vehicle control systems, the hardware and software will need to be exponentially more robust than these "experimental aircraft" computer control systems designed and programmed by amateur and professional hobbyists in their free time.
But the tech is already out there in the wild, and freely available to anyone with internet access. You cannot put that genie back in the bottle. You can restrict access, sure. But ne'er-do-wells have already archived this stuff and will share it between themselves covertly just like they do the
Anarchist Cookbook.
mnem
