Improved Reinforcement Learning Path Planning Algorithm

The topic on coverage path planning (CPP) for multiple cooperative mobile robots using a “partition and cover” approach with Voronoi partitioning to prevent task duplication. By incorporating geodesic-distance-based and Manhattan-distance-based Voronoi generalizations, partition continuity is ensured even in the presence of obstacles, avoiding coverage gaps. The area is divided into <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{2}\mathbf{D} \ast \mathbf{2}\mathbf{D}$</tex> grids, where D represents the robot's footprint size. The proposed partitioning method eliminates coverage gaps and overlaps, as demonstrated by two single-robot CPP techniques: Boustrophedon decomposition and Spanning Tree coverage. Validation is done through simulations and experiments.