Large-scale Topological Environmental Model Based Particle Filters for Mobile Robot Indoor Localization

This paper describes an improved particle filters for mobile robot indoor localization based on a large-scale topological environmental model. The model is composed of multiple pivoted nodes and the connecting relations between them. We only choose the crossings as the nodes in the model so that the indoor environment can be presented by simplest structure. For localization, the traditional particle filters uses large number of samples when applied in large-scale environment, which greatly increase the computational cost and the memory space. To solve this problem, we improve the particle filters by selective distributing. We estimate the possible nodes which the robot is adjacent to in the large-scale topological environment model by laser scan-matching, then distribute the samples around the candidate nodes so that the number will decrease remarkably comparing with the traditional particle filters. Experiments result implemented in real mobile robot platform and further localization result analysis show the validity and practicability of this method.