Path planning based on fluid mechanics for mobile robots using unstructured terrain models

Mobile robots using a 360° field of view LIDAR ranging sensor can generate enormous 3D point clouds. To reduce the quantity of data in memory a compression can lead to unstructured environment models such as irregular meshes. This kind of structure can contain deformed cells and the path planning can be cumbersome. This paper presents a path planning method based on fluid mechanics able to deal with unstructured terrain models. The algorithm uses the finite element method to compute a velocity potential function free from local minima. Then, several streamlines are computed as a road map and the optimal path is selected among the candidate paths. The approach is implemented on the Canadian Space Agency (CSA) Mars Robotics Testbed (MRT) rover and tested at the CSA Mars Emulation Terrain (MET). To confirm the feasibility of the method, the path planner has been tested on 284 LIDAR scans collected in a realistic outdoor challenging terrain.