Rover: Enable driving backwards when using attitude setpoints (offboard)

[jbeyerstedt]

The Offboard attitude control can now drive backwards. The steering angle is reversed to steer in the right direction.

The possibility to drive backwards was not added to velocity control, because we would need to guess the intended driving direction from the data then. Maybe the difference of velocity vector and heading/ yaw data? (Currently heading is ignored, because the "normal" rovers can not move in one direction and look in another.) But I don't see that this is an urgent addition.

[Jaeyoung-Lim]

@jbeyerstedt Are there any logs regarding the tests you did with this? I am curious if there would be any corner cases

[jbeyerstedt]

I have a little test script, that drives in the following way (using MAVSDK offboard attitude control):

• heading 90 deg, throttle 0.2 for 6 seconds
• heading 135 deg, throttle 0.2 for 9 seconds
• heading 90 deg, throttle 0.2 for 9 seconds
• heading 90 deg, throttle -0.2 for 9 seconds
• heading 135 deg, throttle -0.2 for 9 seconds
• stop (heading 90 deg, throttle 0) and disarm

Logfile: https://logs.px4.io/plot_app?log=67261d8f-db40-4978-bf33-3d9a178503c8

[Jaeyoung-Lim]

@jbeyerstedt Awesome script! The last 135 deg in reverse looks a little problematic for me.

Any idea why it didn't track the last yaw setpoint? Also it seems like it didn't have enough time to settle in to the yaw setpoints?

[jbeyerstedt]

You mean at 35 seconds, where the yaw set point moved back to 90 degrees?
At this time, I stopped (throttle 0), but the simulation still "drifted" a bit, because it does not brake when setting throttle to 0. So at least for a car-style rover, the last set point isn't actually reasonable.

I just uploaded my testing script as a gist: https://gist.github.com/jbeyerstedt/9a290a4e87678aee8719632338e8bae4

[Jaeyoung-Lim]

Thanks, looks good on my side

[LorenzMeier]

Nice. I've architected PX4 in 2011 to support reverse trust hoping for this moment. Finally we need it!

[swimmingseeds]

Does the 3rd component of the euler angle ever go negative?

I was thinking about a scenario where the target is more than 180 degrees apart from the rover's front axis 🍔. In that case, would the angle value be negative or go over 180 degrees?

• https://en.wikipedia.org/wiki/Euler_angles#Geometrical_definition

[Jaeyoung-Lim]

@swimmingseeds I think your comments are not related to this PR. Please don't hijack the topic of this PR

If you have any questions, post it on discuss.px4.io or if you have any issues post it as a new issue.