Removing 3D LIDAR 'rings' from pointcloud for ground estimate?

asked 2019-11-19 20:45:43 -0500

PG_GrantDare gravatar image


I am fairly new to pointcloud filtering but have done some research and found that the best method for ground estimate in pointcloud data is by removing the lidar rings. I have run a pcl::SampleConsensusModelSphere model through RANSAC to estimate the ground and am finding that between my data sets occasionally it will detect walls and occasionally the actual ground. The robot will be operating on hilly terrain. Is there perhaps a better model to model these rings, or perhaps there is a way to pre filter the data to remove walls?

All help is appreciated :) thanks.

edit retag flag offensive close merge delete


Would also like to know this

Samira gravatar image Samira  ( 2019-11-21 11:32:01 -0500 )edit

I tried the sphere model but with the curvature of the noise, the walls often fit more points than the ground did. I then looked at the filters here. I ended up trying an adaption of the ray filter but haven't been able to test yet. I would still like to hear how other people approached this.

PG_GrantDare gravatar image PG_GrantDare  ( 2019-11-21 15:58:25 -0500 )edit

Give a try for approximate_progressive_morphological_filter of PCL. This worked for me

kosmastsk gravatar image kosmastsk  ( 2020-06-22 06:54:13 -0500 )edit