To improve the previous versions of Rapidly-exploring Random Trees (RRT) algorithms, a novel algorithm based on Circular Arc Fillet (CAF-RRT*) for path planning problems in two-dimension workspaces is proposed. Firstly,… Click to show full abstract
To improve the previous versions of Rapidly-exploring Random Trees (RRT) algorithms, a novel algorithm based on Circular Arc Fillet (CAF-RRT*) for path planning problems in two-dimension workspaces is proposed. Firstly, it is implemented by combining Quick-RRT* with the bidirectional RRT algorithm, which can obtain an initially planned path. Secondly, the cost of the initial planned path cost is reduced by a proposed path optimization strategy based on the triangle rule. Finally, based on the circular arc fillet method, a path smoothing strategy is proposed to process the middle nodes of the path, while further reducing its path cost. The experimental results on the MATLAB simulation platform show that compared with other RRT algorithms, the proposed algorithm can reduce the path length by about 2%, ensure the smoothness and security of the path, while increasing the running speed of the algorithm by about 95%, and the experimental results of the robot based on Robot Operating System (ROS) platform also prove the advantages of the algorithm.
               
Click one of the above tabs to view related content.