Research on AGV based on improved A-star algorithm

Automated Guided Vehicles (AGVs) are crucial for enhancing efficiency in logistics automation. To address path planning inefficiencies in complex warehouse environments, an improved A-star algorithm is proposed. A three-neighborhood search strategy is introduced, incorporating obstacle detection and dynamic direction adjustment to eliminate redundant node traversal. Additionally, an optimized evaluation function is developed by integrating a predictive cost weighting coefficient based on OPEN list. Comparative simulations on large-scale maps with varying obstacle densities demonstrate the algorithm's superiority and robustness. Results indicate that compared to Dijkstra and traditional A-star, the proposed method reduces search node traversal by up to 95.9% and 37.6%, respectively, and computation time by 94.8% while maintaining optimal path length. The algorithm exhibits consistent performance advantages across different environmental complexities, validating its scalability and reliability for real-time logistics applications. Furthermore, comparative analysis with state-of-the-art planners (e.g., JPS, Theta*) highlights its practical advantage of balancing high performance with implementation simplicity within the standard A-star framework, ensuring easy integration for real-world AGV systems.