Revision history [back]

From the edges of the walls in that photo, it looks to me like you've got some pretty messy data coming in. That messy data might be due to to the LIDAR: sensor noise causing obstacles to be seen where there are none. Or, that messy data might be the odometry: if the odometry stack makes the turtlebot think that it's in the wrong place, it might be seeing a valid obstacle and projecting it to the wrong location.

In either case, it looks like you'd benefit from some filtering. For your LIDAR, check out the laser filters package. For your odometry, the more data sources you can take advantage of the better. Combining them alongside some filtering using some form of a kalman filter would help, too (perhaps via this package?).