Abstract:

To improve the efficiency and quality of dynamic path planning for mobile robots in complex environments, this paper proposed a path planning algorithm that combined an improved RRT-connect with the DWA. Two expanding random trees were introduced for alternating expansion, and a dynamically restricted sampling area was set to reduce the randomness of the sampling process while ensuring the probability completeness of the algorithm. Atarget bias adaptive step size strategywas employed to enhance the target orientation of the random tree expansion process. A greedy strategy was adopted to prune redundant nodes in the path and smooth the global path to obtain the globally optimized path. The DWA was employed to track the global path, while the evaluation function was improved by introducing a path tracking evaluation function with an adaptive weight strategy. At the same time, a path correction mechanism was introduced to remove invalid local target points and avoid the phenomenon of backtracking. Simulation results demonstrate that compared with the previous method, the proposed algorithm reduces running time by 36.18%, 68.61%, and 89.33% and shortens path length by 17.61%, 17.48%, and 12.33% in different static environments. In dynamic environments, it reduces running time by 31.46% and shortens path length by 9.21%, while consistently ensuring path safety.

Key words: mobile robot, adaptive, RRT-connect, dynamic sampling area, DWA, path planning