Closed piyushk closed 11 years ago
I take back the statement that it was not affecting navigation.
I was running into a problem with the global planner not re-planning correctly if an unexpected obstacle snapped the band. I introduced a hack in https://github.com/utexas-bwi/segbot_apps/commit/4e3025dd26ce4551d0a4ae8f9b3614887f0e9424 that would wait for the global planer to catch up. As it turns out supply the unfiltered topic to the global planner was causing this issue, though I don't know why. I'll revert the commit.
Never mind. I think fixing the topic helped, but I was finally able to reproduce the issue after reverting https://github.com/utexas-bwi/segbot_apps/commit/4e3025dd26ce4551d0a4ae8f9b3614887f0e9424. No harm in leaving it in.
The costmap has been listening to the incorrect topic:
For navigation, scan_filtered should always be used. The hokuyo does produce scan, but it contains points from the robot's body
This was not affecting navigation too much, as the costmap ignores readings from within the robot's footprint. However it would not have worked with the kinect configuration.