A novel qualitative motion model based probabilistic indoor global localization method

This paper presents a novel global localization approach based on a qualitative motion model to address the weakness in traditional probabilistic localization methods. To represent the occupancy global grid map for efficient localization, we generate a global hypothetical pose set on the grid map. Then the map is represented as a set of virtual observations associated to the poses. The online global localization is implemented based on the proposed enhanced particle filter. First, the particle set is obtained by a sampling process in the generated candidate hypotheses set where the robot may be located. Second, the particle set is propagated using a qualitative motion model and the negative effect of the motion model uncertainty are eliminated. Third, in the particle tracking process, the historical information and current observation are incorporated in the belief of the particles. The robot pose is estimated as the particle with the highest weight. In addition, the localization accuracy can be adjusted by sampling hypothetical poses with different densities. The results of the experiments show that the proposed method is robust to motion model uncertainty and can perform accurate global localization.