Need a 3rd steering control type

[Gernby]

While working through the latcontrol issues that some Toyota's experience, I noticed that almost all cars have yaw sensor data defined in their DBC's (not Honda), but it seems that all OP-compatible cars will have it available (like my Honda Accord). After researching the value of yaw rate in modern vehicles for ADAS systems, I really think that there needs to be a third control type added for steering that is based on actual / desired yaw rate.

Obviously, angle-type steering simplifies the control logic greatly. Also, torque-type steering works well for straight-rate vehicles that have high resolution steer-rate data on the CAN bus. However, for all the other vehicles, there's too much error to really make it work well.

With a new yaw-type steering classification, I think that "problem cars", like the Toyota Prius Prime can be made as ideal as any other. The reason why I think this will be better is because it won't rely so much on the steering angle sensor, which seems to introduce a ton of error. It would also eliminate error caused by variable steering rates, and reduce lag caused by sticky / heavy / springy steering columns.

I've already started working on this, but won't have it ready for a few days.

[Gernby]

I didn't mention it in the first post, but after reviewing my dashboard closer, I really think that the Honda Accord might not use a torque based request at all.

When OP's desired steering angle is converted to yaw rate (VM.yaw_rate(self.angle_steers_des * CV.DEG_TO_RAD, v_ego)), the plot of it compared to the commanded torque has very strong resemblence. The green line in the plot below represents the calculated yaw rate from the MPC's desired angle, and the stacked plots of yellow (i), blue (f), and red (p) represent the torque command..

Note that this is not the first time I doubted the Accord was torque-type. I theorized many months ago that the Accord might actually use angle-type steering commands, based on observations at lower speeds. However, that was quickly disproved at higher speeds, which caused violent steering response.

My new theory that the Accord is based on Yaw Rate seems to explain my prior observations.

[pd0wm]

I assume this issue is now solved by the new INDI controller.