Robust localisation for mobile robots in indoor environments

Abstract

In order to operate reliably in industrial environments, a mobile robot must know where it is. As part of a student project, a method has been developed for robust localisation in a partly ambiguous environment using a particle filter. This enables accurate estimations of the robot's position by merging odometry readings with measurements provided by a 240 degree laser rangefinder. The developed algorithm is able to solve the kidnapped robot problem in realtime with feasible computational effort. Even though the presented approach is customized for the given setup it can be easily adapted to perform in other environments.