AGV Path Planning Algorithm Based on Fusion of Improved A* and DWA

Aiming at the problems of excessive turns, poor path smoothness and oblique crossing of obstacle vertices in the traditional A* algorithm, this paper optimizes the traditional A* algorithm and combines with the improved DWA algorithm for real-time obstacle avoidance for $A G V$ path planning. To optimize the $A^{*}$ algorithm, firstly, the evaluation function is improved by adding dynamic parameters. Secondly, the bidirectional path-finding method is adopted. Then, this paper designs a key node extraction strategy to remove redundant points. Finally, the Bezier Curve is used at the inflection point to make the route smoother. The DWA algorithm with improved evaluation function is used between the key nodes obtained by the A* algorithm. The simulation results show that the improved $A^{*}$ algorithm is superior to traditional $A^{*}$ algorithm in terms of path length, number of inflection points, path smoothness and number of traversal nodes. The improved DWA algorithm enhances the obstacle avoidance ability and is less likely to fall into a local optimum over traditional DWA algorithm. The fusion algorithm makes AGV reach the target node on the basis of global path planning, and avoid obstacles in real time to achieve local optimum.