Entropy Gradient-based Exploration with Cooperative Robots in 3-D Mapping Missions

Building cooperatively 3-D maps of unknown environments is one of the application fields of multi-robot systems. This article focuses on the exploration problem with multiple robots, starting upon a previously proposed successful distributed architecture. An entropy gradient-based algorithm is used to select exploration viewpoints in the frontier between explored and unexplored regions. The architecture is refined here with a mutual information-based coordination mechanism, whereby each robot selects new exploration viewpoints so that mutual information between robots’ individual visible maps is minimized and map’s uncertainty is decreased as fast as possible. Results obtained from experiments with real robots equipped with stereo-vision demonstrate how the entropy gradient-based method converges to a map with lower uncertainty.