I have been running a lot of laps in debug trying to work on getting a narrow tyre to perform. Whilst doing this I have observed that we still get a lot of 'spiking' in the steer torque figure (up to 180Nm when the mean is probably around 30Nm) indicated in the middle of the debug screen even in fairly constant rate turns - I assume this is generated from the tyre model for some reason.
However, this spike appears to preceede the virtual rider over-reacting and forcing the bike over (a sort of capsize mode) - I have observed this behaviour over many betas but this is this first time I have had steer torque indication to correlate to.
Are any of the rider PID values able to apply a non-linear damping function (ie f(x)=X sin(torque) ? i.e. to supress the very brief extremes (its maybe only 1 sample) but maintain sensitivity to the mean and allow the rider to act on the more realistic time frame of the averaged input?