Revision history [back]

To remove far points, set a max range for the local occpuancy grids. To fill unknown space, we need to enable ray tracing (which is enabled by default using a lidar). Here are two approaches:

2D approach: faster/less memory, though obstacles may be cleared wrongly if they don't appear in the field of view of the camera

Grid/RangeMax=3
Grid/RayTracing=true
Grid/3D=false
Grid/MaxObstacleHeight=2

3D approach: slower/more memory needed, ray tracing is done in 3D, so it is safer for obstacles that camera don't see anymore

Grid/RangeMax=3
Grid/RayTracing=true
Grid/3D=true
Grid/MaxObstacleHeight=2

If you can assume that all obstacles in front of the robot can always be seen in the field of view of camera, I would go for the 2D case for cpu/memory efficiency.