Research on dynamic local path planning for robots based on centerline extraction algorithm with improved DWA
Junyan Qian, Zihao Wang, Bin Yan · 2024
With the increasing complexity of dynamic environments, the challenges faced by traditional DWA algorithms in path planning have become more and more prominent, and the accuracy of the algorithm pairs needs to be improved in terms of real-time path planning capability. At the same time, the algorithm is often inefficient due to high computational complexity and difficult to run on embedded chips. To this end, this paper proposes an improved DWA algorithm, which improves the path finding ability by optimizing the angle evaluation function and introducing a local dynamic target point mechanism, improves the algorithm efficiency by improving the obstacle evaluation function and introducing a hybrid mechanism, and introduces an artificial potential field to solve the problem of obstacle avoidance for consecutive obstacles, so as to improve the algorithm's generalization ability. Compared to the traditional DWA algorithm, the proposed algorithm results in an intelligent vehicle with excellent trajectory and a 24.44% reduction in cruising time on the same track. The frame loss due to computation is also reduced and the frame drop rate is reduced by 76.48%. The algorithm proposed in this paper is suitable for local path planning scenarios of intelligent vehicles for continuous obstacles such as road edges, and can promote the development of intelligent driving path planning on real roads.