Weihua Xiong, Xiaoyun Chen, and Ruisong Ji
A Algorithm, dynamic window approach, b´ezier curve, K-dimensional tree, obstacle avoidance, global–local integration
This paper proposes an improved integrated global and local path-planning algorithm to overcome the major limitations of the traditional A algorithm in automated guided vehicle (AGV) applications, such as low search efficiency, excessive node expansion, and inadequate adaptability to dynamic environments. The proposed method incorporates an environment evaluation mechanism to adaptively select a forward, backward, or bidirectional search strategy. It also integrates a node-filtering scheme with safety distance constraints and an obstacle penalty factor to enhance path safety, and applies a dynamically adjusted B´ezier curve for global path smoothing. For local navigation, the algorithm is combined with the dynamic window approach (DWA) and utilises a K-dimensional tree (KD-tree) to ensure consistency between local trajectories and the global plan. Two enhanced variants of A are implemented as baseline methods: a dynamically weighted heuristic approach with secondary optimisation, and a bidirectional search method with an improved strategy. Simulation results in complex logistics scenarios demonstrate that the proposed approach improves search efficiency by 40%, reduces expanded nodes by 30%, and decreases the maximum path deviation by 60%, showing clear advantages in efficiency, smoothness, and safety.
Important Links:
Go Back
