pybricks
[Feature] Also reset heading target when resetting odometry.
Is your feature request related to a problem? Please describe.
At the moment, calling drivebase.reset() resets the odometry offsets, but doesn't do anything to the controls. So if you are holding, you will keep holding.
Based on various discussions like https://github.com/orgs/pybricks/discussions/1616#discussioncomment-9251463, I think the user assumption is that this will also reset any internal controller states.
Describe the solution you'd like
I think it is reasonable to reset the states equivalent to calling stop(), so the user does not have to do this themselves, and so they don't forget about it.
drivebase.reset() is usually used when re-aligning the robot with the wall, so having a clean slate here makes sense, so it won't keep trying to maintain any previous targets in case it was holding.
I think it may also be reasonable to assume that the user will want to stop when calling drivebase.reset(), so maybe physically coasting isn't a bad idea. You could always immediately hold again if you wanted to.
When we do that, reset really does mean: reset back to the start, as if starting a new mission/program, which is probably what this function is going to be used for anyway.
This is a slightly breaking change though, so we'll want to clarify that in the documentation.
Sounds good and awesome. My dream drivebase is global-orientation aware, therefore I would still be happy to see an optional parameter to use for the heading initial state.
robot.reset(angle: Number = 0)
This allows an FLL team to use reset between two missions still keeping the orientation global (N-W-S-E) no matter how they position the robot. FLL and WRO mats have the orientation N facing top and S on the bottom and so forth... In some runs the bot starts facing north, while in others it faces west or east or even south.
This if the team starts the bot facing leftwards it would call robot.reset(-90) instead of the default robot.reset() thus robot.angle() still yielding -90 when facing to the west.
For Python, the method call as described by @afarago sounds good to me. This would also mean that there is no way to only reset the distance.
As my teams are using Block Coding, the name of the block "reset distance and angle" and the example given in https://www.patreon.com/posts/pybricks-3-5-is-102449785 is a bit misleading when you enable useGyro, if you don't know, that the actual reset of the heading is done by the "drive base stop coast" block. If you used "drive base stop HOLD" there, you get the behavior demonstrated in this video:
https://photos.app.goo.gl/jdEdbWDaD6GCa9B99
So yes, my expectation is also that "reset distance and angle" would reset the heading-angle.
Agree 100% - elementary age group will not handle global heading.
For block coding it would be so nice to have "expandable" blocks - gradually adding more (and more) params to set.
I think the course of action here is:
Change .reset() to .reset(angle=0, distance=0)
Both angle and distance can be worth resetting. As suggested by @afarago.
Decide what to do about maintaining hold mode
We can either include a .stop(). Or, maintain the newly reset angle.
Both should avoid the sitation in @ggramlich's video. Calling .stop() would be simpler but not entirely technically correct.
Augment and document the block This is currently missing from the docs. Potentially add the above options for angle and distance.
Decide relation with hub.imu.heading()
This isn't decided yet. Should it also be reset? It is technically independent... But in practice, it is probably used synonymously with the drivebase angle. There is also the other way around. What should the drivebase do if you call hub.imu.reset_heading(0)? This came up in https://github.com/pybricks/support/issues/1818
EDIT: We should probably make them always the same. This is the easiest to explain. Even if it isn't the "right" approach from how how the firmware is structured, we can still do it.
For these to be proper replacements for imu.heading(), these probably need to become floating point values too.
The fix is available for testing using these instructions to try the latest version
Please re-open if you experience any further issues or inconsistencies. Thank you!