Algorithm or package for road (ground) surface detection from 3 LiDAR for natural areas

asked 2019-11-27 08:38:02 -0600

twdragon gravatar image

updated 2019-11-27 08:38:32 -0600

I am trying to implement road surface detection for mobile robot. Before implementing something by myself (like this: https://ieeexplore.ieee.org/abstract/...), I want to ask the contributors: is there any package or pre-implemented reliable road surface detection algorithm which accepts 3D LiDAR point clouds as input? Especially I am interested in algorithms for natural areas including forest, fields, snow surfaces etc.

edit retag flag offensive close merge delete

Comments

I have been using hdl_graph_slam for localization, but it has a floor detection nodelet. I have not used the floor detection nodelet much (just turned it on to see what it would do), but it may give you a starting point.

rittersb gravatar image rittersb  ( 2019-11-27 09:39:09 -0600 )edit

@rittersb I tried to use RANSAC implementation like in hdl_graph_slam without significant effect. I thought RANSAC is practically not suitable for natural environment with grass especially.

twdragon gravatar image twdragon  ( 2019-11-28 02:08:04 -0600 )edit