A Multi-Robot Mapping Approach using Different Motion Planning Methods

L. Muñoz-Gomez, M. Alencastre-Miranda, and R. Swain-Oropeza (Mexico)


Mobile robotics, motion planning, multi-robot mapping, segmentation.


This paper presents a mapping system with multiple robots for indoor environment. Each robot uses its own laser range finder (LRF) as sensor and builds a local map (segment-based), these maps are sent to a server that joints the partial maps and then generates a global map. The robots relative positions are known in advance at the beginning; during the mapping process each robot has its own information without knowing about each other. The next possible position that each robot chooses for taking a new reading, is selected considering the free edges obtained from the segmented range points. Each time a new reading is taken, segments are matched using the Hausdorff distance in order to find two or more segments that can be fused in a single one. Two different motion planning approaches for reaching the next position during the exploration have been implemented: Markov Decision Process (MDP) and Roadmaps considering part of the environment already known (Visibility roadmap, probabilistic roadmap and visibility probabilistic roadmap). Results using simulations and real robots are presented.

Important Links:

Go Back