ROS Resources: Documentation | Support | Discussion Forum | Index | Service Status | ros @ Robotics Stack Exchange
Ask Your Question
0

Odometry - 4ws quasiomnidirectional - covariance_matrix

asked 2020-02-02 11:41:01 -0600

Dragonslayer gravatar image

Hallo, I had my 4ws base-controller invert wheelspeed if steerangle was bigger then Pi/2 and adjust the steerangle of course. It should improve stability of the robot as well as be more efficient.

However my Odometry inverted reality when the robot was in a state where this function didnt execute for all four wheels. I asume it has to do with the covariance_matrix but Iam totally new to the covariance_matrix topic. Where can I find documentation etc. that might lead to a solution (the usual google search stuff I checked)? Or does anybody know if this might even work at all with a covariance matrix, and what would be the drawback to kicking out the covariance_matrix completly? Thanks.

edit retag flag offensive close merge delete

1 Answer

Sort by ยป oldest newest most voted
0

answered 2020-02-05 13:44:43 -0600

Dragonslayer gravatar image

SOLVED: Was an ugly mixup in the microcontrollers PID/drivers/serialinterface. For anybody else encountering such an issue I would suggest, measure the wheelspeeds/steerangles exact (not by ros but by external means). The right/left wheels were not doing what they should have. Because of friction, weight distribution and steeringangle, the base seemingly did what it should (although miserably bad, but I thought it was because of imperfect tuning). But in reality it was pure coincidence, at twist-Z the coincidence ended, that was the issue. Once more a case of changing the perspective, test setup, and tools was the solution.

edit flag offensive delete link more

Question Tools

1 follower

Stats

Asked: 2020-02-02 11:41:01 -0600

Seen: 160 times

Last updated: Feb 05 '20