search

at-wat / mcl_3dl

A ROS node to perform a probabilistic 3-D/6-DOF localization system for mobile robots with 3-D LIDAR(s). It implements pointcloud based Monte Carlo localization that uses a reference pointcloud as a map.
BSD 3-Clause "New" or "Revised" License
490 stars 117 forks source link

Support matching with intensity #187

Open at-wat opened 6 years ago

DaikiMaekawa commented 6 years ago

looks interesting. you mean, considering intensity a part of matching score in the measurement update, right?

at-wat commented 6 years ago

Most complex part is treating incidence angle of the beam into the surface and reflection characteristics of it. Simply taking intensity data into the likelihood calculation makes them improper. I have some basic ideas to treat them with minimal run-time computation cost.

I'm planning to do it after https://github.com/at-wat/mcl_3dl/pull/195.

greymfm commented 8 months ago

This could be tricky (treating incidence angle). However, depending on the material, one will get a high intensity (white material), a low (black material) and something in between (all other). The high and the low is less dependant on the incidence angle?

Idea: Take the difference multiplied by some small weight? (measurement value - expected value) * weight Then this could at least filter out particles with wrong material (white vs. black or vice versa).