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
480 stars 119 forks source link

Reference for motion prediction model #276

Closed tuandle closed 4 years ago

tuandle commented 4 years ago

Hi @at-wat , Thank you for your work. I have a question about the implemented motion model in this package. The implementation of the diff drive model here seems to be a bit different from what described in Thrun's book and the original ROS navigation pkg.
I wonder if you can provide some references to how you derive formulas in functions setOdoms and predict? I am also confused about the usage of odom error constants, e.g odom_err_integ_lin_tc_, odom_err_integ_ang_tc_ , noise_ll_, noise_la, etc., can you explain what they are for? Thanks for your help!

at-wat commented 4 years ago

The implementation of the diff drive model here seems to be a bit different from what described in Thrun's book and the original ROS navigation pkg.

I don't have specific reference for it. Basically, only thing the Bayesian filter requires is that the ground truth state keeps having enough probability, so you don't need to 100% follow the book. Basic things predict and setOdoms doing are updating the pose of each particle by using difference of the odometry pose with noise specified by the parameters.

I am also confused about the usage of odom error constants, e.g odom_err_integ_lintc, odom_err_integ_angtc , noisell, noise_la, etc., can you explain what they are for?

odom_err_integ_* means: https://github.com/at-wat/mcl_3dl/pull/99#issuecomment-363401362

https://github.com/at-wat/mcl_3dl/blob/15f4330b2174d30cabe75d655c6eba61620e05ad/src/mcl_3dl.cpp#L1219-L1222 These odom_err_*_* have basically same meaning as amcl's odom_alpha3, odom_alpha2, odom_alpha4, and odom_alpha1 respectively. (http://wiki.ros.org/amcl#Parameters#line-395-2)

tuandle commented 4 years ago

@at-wat I got it! Thanks for your reply!