or-tal-robotics / mcl_pi_gazebo

Multirobot cooperative localization using the Particles Intersection algorithm

mcl_pi_gazebo

By Or Tslil, Tal Feiner

Particles Intersection (PI) is a method to fuse multiple particle filter estimators, where the cross-dependencies of the observations is unknown. This package contain a PI implembtation for fusion the particle filters that estimates the state of a robot in a flat space. This package contain two main nodes:

• particle_filter.py - an implemantation of a particle filter for robot localization using [1].
• particlesintersection.py - an implemantation of a particle intersection for cooperative localization using [2].

Example

Dependencies

The following python packges are required:

• python 2.*
• numpy
• matplotlib
• sklearn
• sciPy
• teb_local_planner for move_base

Runing

For a simulation with 3 robots and move_base controller use:

roslaunch mcl_pi_move_base.launch

For a remote control (needs 3 controllers) use:

roslaunch mcl_pi.launch

References

[1] Thrun S, Burgard W, Fox D. Probabilistic robotics. MIT press; 2005 Aug 19.

[2] Tslil, Or, and Avishy Carmi. "Information fusion using particles intersection." 2018 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 2018.