Development of a new technique in ROS for mobile robots localization in known-based 2D environments

Abstract EN

Adaptive Monte Carlo Localization (amcl) is the only standard package for mobile robots localization in Robot Operating System (ROS).

In this research, a new particle filter based localization technique named general Monte Carlo Localization (gmcl) was developed by adding three particle filter algorithms to amcl in order to improve its performance, so the new versions of ROS could be better invested in systems that depend on the knowledge of the robot’s pose.

In addition, we compared amcl and gmcl in terms of computational complexity and the ability of addressing the pose-estimation problem in a differential drive mobile robot equipped with a LiDAR sensor.

The results showed that the new proposed technique outperformed amcl in the accuracy of estimating the pose when compared to the same maximal computational workload.

gmcl was able to reduce the pose-error in pose tracking and also able to increase the success rate of robot’s pose detection in the two problems of global localization and kidnapped-robot.