Robot path planning with multiresolution probabilistic representations: a comparative study

Path planning is an essential step in robotic applications; it allows the robot to execute a task successfully, without any collisions with neighbouring objects. As path planning performances directly depend on the nature and the encoding of an environment representation, various approaches can be considered. New strategies are studied and compared with classical ones: multiresolution versus standard occupancy grids and probabilistic versus deterministic datasets. The goal is to identify the best solutions for path planning and collision avoidance in semi-autonomous robotic systems operating in complex environments.