Yaw drift in narrow areas about als_ros HOT 2 CLOSED

Have you ever heard that procrastination can be a virtue at times? Because a proper delay gives others time to meditate and prepare, and give yourself time for other pressing tasks.

Here it is again. I think I found the solution by plugging away at the problem.
Actually there is some hint from the jittering in the above video.
The cause is missing velocity from the odometry messages at /odom topic.
Now with the velocity info, the als_ros outperforms amcl.
Cheers.

from als_ros.

After thoroughly testing als_ros vs amcl and many tentative parameter tuning, I found that the als_ros is much more brittle than amcl. Though it updates at the rate the lidar scans, it often drifts and then unable to recover. When the global pose sampler is enabled, the estimated poses often jumps between candidate poses. The two issues do not exist with amcl in ros1. In the end, I turn back to amcl to complete the project.

from als_ros.