Hello,
I can't figure out how to increase the weight of the odometry against the SLAM relocalization when building a map with gmapping.
As I am building a map in an environment with featureless long corridors with a limited range LIDAR (4m), I am trying to avoid jumps in the map building. My odometry accuracy is pretty good, especially in translation and if i replay the lidar data in rviz with a long decay time I can see the corridor drawing nicely at the correct dimension. However the corridors are way too short in the map build. This is a well know issue of environments with low feature and one solution is to tell the algorithm to trust more the odometry. I have set minimumScore to 100 (no more cause I still want to detect loop closure and to compensate rotational drift), and i have play with various value of srr and srt (even 0!) but changing these values seem to have absolutely no effect.
So my question is, short version: how to increase the weight of odometry when building a map with gmapping?
Long version: what is the impact of srr and srt, and what they stand for?
Thanks!
↧