Information-based Active SLAM via Topological Feature Graphs

Active SLAM is the task of actively planning robot paths while simultaneously building a map and localizing within. It is challenging in that the feasibility of paths will depend on future observations along the path, and the complexity of the problem grows quickly with the size of the map and the length of robot trajectory. This work proposes a Topological Feature Graph (TFG) representation of the map that scales well and develops an active SLAM algorithm with it. The TFG enables a unified quantification of exploration and exploitation gains with a single entropy metric and hence facilitates a natural and principled balance between map exploration and refinement. A probabilistic roadmap path-planner is used to generate robot paths in real time. Results from hardware experiment show that the proposed approach achieves better accuracy than the traditional grid-map based approaches while requiring orders of magnitude less computation and memory resources.