While what you’re writing and thinking is interesting, it seems to be missing the practical reality of what onewheels (and VESC onewheels) generally see in their daily life.
There are reasons why this board shape has foot sensors, in one form or another.
On the IMU specifically, there have been, and continue to be, times when IMU confusion leads to odd balancing or ride behavior. This can be from slightly bad calibration, a bad IMU, brown outs on the 5v/3.3v rail for one reason or another, a slightly untuned zero vector frequency (currently, a massive thorn in the side of people tuning VESC onewheels and their ride feel), and probably other things.
Relying on the IMU to detect the rider would lead to any IMU confusion or issues to risk a runaway/ghosting board. The same would happen if the IMU simply didn’t register the conditions for dismount.
Ghosting already happens for a number of reasons to do with the footpad, or the tuning of a VESC onewheel, and whenever it does, it’s very scary. A sensor can be damaged or defective, or cold temperatures cause the plastic layers to rebound too slowly to open the switch, or a defective sensor leads a user to disable moving faults, and they encounter a ride condition where that leads to a jump off and a ghosting situation.
Generally, that’s kind of rare, but I think it’s rare because the foot sensor has remained a physical switch that, so long as it’s checked to be working, will operate reliably whether or not the IMU is functioning and calibrated as expected.
Even with a bad IMU, a rider can feel a problem and jump off, and the board will disengage.
I’m not sure if you remember or paid attention to the launch of the Onewheel GT, but due to both the sensor changes and the firmware conditions for engagement, many boards ghosted and led to property damage and injury. One NYC rider had his board ghost backwards and break a pedestrian’s ankle.
That’s all to say, that currently, the way foot sensors are (a physical switch), seems to be on the side of more consistent activation/deactivation so long as when setting up a VESC onewheel, the builder can verify that the sensor actually works and doesn’t show any delay of the ADC voltage changes.
Even outside of IMU confusion, and circling back to the general life of a onewheel, there are many conditions of riding that can and do subject the board to weird, jarring, and impactful movements that can appear almost exactly like a person getting onto the board. Trick riders manipulate the board while on and off, trail riders have their boards tumble down cliffs and bounce around during crashes in odd ways, and onewheels are generally expected to take large amounts of physical abuse and continue to work normally, VESC onewheels included.
The idea of software based foot sensing is interesing, and any new ideas are always interesting.
Stoked Stock and Tech Rails have been working on a LiDAR based foot sensor, and that seems novel and interesting. But, it still leans on the sensor itself being a separate and purpose-focused thing that only concerns itself with whether or not a rider is on the board. And I think that’s because rider detection is near the top of safety concerns for a onewheel.
Riders do fall, and that’s always bad. However ghosting boards are a larger concern, as they can (and have) injured bystanders who didn’t ask for it or have anything to do with the ride.
Mixing rider detection with components that do other things on the onewheel, I generally remain skeptical of.
Any suggestions of how that could be detected without footpads (except pushing the turn-off button)? Some creative thinking is required. This state might be the reason why footpads are required.