Mobile Robot Path Planning Algorithm Based on Improved RRT*FN

Abstract

For the problem that the rapidly-exploring random trees star fixed nodes(RRT*FN) algorithms, with a fixed number of nodes have large randomness and low search efficiency. An improved RRT*FN path planning algorithm for mobile robots is proposed. First, the improved algorithm inherits the advantages of RRT*FN in optimizing memory and introduces a heuristic sampling strategy to add high-performance nodes to replace inefficient ones when the total number of nodes in the tree reaches a preset value. Secondly, an adaptive extension strategy is used to automatically adjust the weights of tree growth toward random and target points by collision detection methods. Finally, it is demonstrated through simulation experiments that the improved algorithm takes into account the exploration of obstacle-dense and narrow regions while fixing the number of nodes and optimizing the memory. The improved algorithm is more efficient in the path planning process and has stronger environmental adaptability.