Objective In the context of path planning for automated guided vehicles AGVs in warehouses the traditional A* algorithm suffers from several issues. These include a high number of node expansions excessive turning points and paths that cut across obstacle vertices which can easily lead to collisions in real-world environments. To address these problems an improved A* algorithm is proposed. Methods First to prevent the planned path from passing through obstacle vertices and thus reduce collision risks the search direction for expandable neighbor nodes was modified when obstacles were present in the neighborhood. The heuristic function was dynamically adjusted through the combination of two factors the obstacle ratio between the current node and the target node and the distance ratio of the current-to-target segment to the start-to-target segment. After the initial path was generated redundant nodes were removed using a polyline optimization method and the path was smoothed by fitting it with an improved quadratic Bezier curve. Results MATLAB simulation results demonstrated that the improved A* algorithm effectively avoided obstacle contact achieved fewer turns