3 ms·
Is this any better than a Kalman filter or just a lot more computationally intensive?
by colechristensen 5y ago
Is this any better than a Kalman filter or just a lot more computationally intensive?
- advisedwang 5y agoKalman filter solves the problem of getting accuracy from noisy or infrequent speed data. However with only acceleration data you still can have integration errors build up over time. IN other words, Kalman filter can't solve that the driving steadily at 50 and 55mph are effectively indistinguishable from acceleration data. OTOH I'm skeptical that this method really solves it either
- sydthrowaway 5y agoThis. Basically, fuck (MEMS) accelerometers. I struggled with this in my early university days. "Wait, we get acceleration? Why not just cumsum twice!" except get garbage. So this is a fuck you to accelerometers for wasting my time
- sokoloff 5y agoI’d expect you could get some signal of speed from the periodic double-thumps of expansion joints, speed bumps, potholes, and the like. Combine that with acceleration integration from stops to find an estimate of the wheelbase. It might not be practical or the easiest way, but I suspect that you could tell the difference between steady highway driving at 50 vs 55 from acceleration data only (due to inherent road imperfections).
- advisedwang 5y agoThat's really clever! That's also the kind of thing an AI could pick up on, so maybe this is a stronger technique than I initially thought. Although autocorrelation is probably a better way to get that data than just throwing it into a AI model.
- csours 5y agoI wonder if that generalizes to other kinds of vehicles or even other instances of the same kind of vehicle.