If you use PID to stabilize an unstable system, is that in the top-right corner of the cartoon map? I suppose there are implicit "prediction bands" also with standard PID control in the sense that the gain and phase margins have to be sufficient. The plant has to exist within a certain range. There is no need to do it, but I guess it is possible to visualize a possible range of open loop short time ahead forecasts based on the allowed plant range, from the at-present state. Bit contrived maybe.
Ben, one question regarding the work of Shalev-Shwartz. The tweet you linked is much more concise than what you describe. Do you have a reference for that handy? Thank you.
Hi Miguel, I don't have a reference for what Shai is specifically referring to. I pooled what was in his tweet and what I personally know about autonomous driving.
If you use PID to stabilize an unstable system, is that in the top-right corner of the cartoon map? I suppose there are implicit "prediction bands" also with standard PID control in the sense that the gain and phase margins have to be sufficient. The plant has to exist within a certain range. There is no need to do it, but I guess it is possible to visualize a possible range of open loop short time ahead forecasts based on the allowed plant range, from the at-present state. Bit contrived maybe.
You are ahead of me by about a week. :) I'm hoping to talk about robustness and margins in SISO feedback loops soon.
Ben, one question regarding the work of Shalev-Shwartz. The tweet you linked is much more concise than what you describe. Do you have a reference for that handy? Thank you.
Hi Miguel, I don't have a reference for what Shai is specifically referring to. I pooled what was in his tweet and what I personally know about autonomous driving.
No worries Ben. The comment about tracking kinematic states of relevant objects/entities was so specific that I thought you were quoting a paper.