Normalize drone orientation errors to [-pi, pi] - #218
Conversation
|
This can be merged as soon as the issue on MacOS 11 is resolved. Until that point, we cannot run the minimal set of tests that were defined and I do not think we should merge untested changes into the master branch. |
|
@ilpincy the issue seem to be related to the tcl-tk dependency |
|
The test scripts of Github Actions are broken at the VM setup phase. This code does not affect those tests. I reviewed the code changes. The original code was completely wrong, so this changes are a welcome addition. The changes, however, only consider the Z axis despite being committed to a 3D engine. Still better than the previous version, so I'll merge them. Thanks for the contribution! :-) |
|
Hi, Michael, |
…cal movement and rotation
|
Hi, Carlo, Michael, I made another commit about qtopengl for you to consider, this may should belong to another PR, but I can't create a new one with the current one existing, so I made a commit here. The check failure about MacOS still seems to be there, and I saw another check failure saying can't find qt5-default package. Is it my fault? Let me know how I can do things more properly. This commit consists two aspects:
Weixu |
Problem description:
When using drones, first set
robot.flight_system.set_target_pose(vector3(0,0,0), math.pi - 0.01)
and then set
robot.flight_system.set_target_pose(vector3(0,0,0), math.pi + 0.01)
instead of keep rotating, the drone tends to rotate back through the longer side to math.pi+0.1
I tweaked the Z component of orientation error and angular velocity error in drone pointmass model to make the drone rotates smoothly when jumping across the edge of Pi.
A small test scenario is enclosed
vns.zip