LaCAM*-SFDWA: Multi-autonomous mobile robots’ path planning in warehouse environments based on spatio-temporal fuzzy dynamic window approach

To address the challenges of multi-autonomous mobile robots’ path planning in warehouse environments regarding dynamic obstacle avoidance and parameter adaptability, a hierarchical path planning framework based on the lazy constraints addition search and spatio-temporal fuzzy dynamic window approach is proposed. At the global planning layer, the LaCAM* algorithm is employed to generate collision-free, globally optimal paths for multi-autonomous mobile robots; at the local trajectory planning layer, a novel spatio-temporal fuzzy dynamic window approach algorithm is introduced. First, a spatio-temporal dynamic avoidance model based on the velocity obstacle principle is presented, constructing a risk cost function that is subsequently embedded into the dynamic window approach evaluation function to enhance real-time collision avoidance capabilities. Second, a multi-level fuzzy logic control system is designed to adaptively adjust evaluation weights, overcoming the limitations of fixed coefficients in changing environments. Furthermore, a local optima escape strategy is developed, which guides the robot out of local optima by replanning to generate temporary subgoal points. Experiments demonstrate that the framework generates smooth, safe trajectories, improving navigation stability and dynamic obstacle avoidance efficiency for multi-autonomous mobile robots.