Answers are grounded in this paper and cite their sources.
(1) Background: Multi-AGV path planning in narrow warehouse environments remains challenging due to dynamic obstacles, frequent inter-AGV conflicts, and the difficulty of balancing safety and efficiency. Conventional A*-DWA frameworks typically rely on static global paths, reactive local avoidance, and fixed evaluation weights, which limits their adaptability in dynamic and congested environments. (2) Methods: To address these issues, this paper proposes a risk-aware planning framework, termed RPC-A*-DWA, which integrates an improved A* global planner, short-term dynamic obstacle prediction, composite risk assessment, risk-adaptive DWA, and distributed priority-based conflict resolution. The framework uses distance, time-to-collision (TTC), and local congestion to quantify environmental risk and dynamically adjusts the local planning strategy accordingly. It is evaluated in single-AGV tests, ten-AGV simulations, an ablation study, and real-robot trials. (3) Results: Compared with the static A* baseline, the proposed method reduces peak risk by 67.3% in the single-AGV scenario; compared with the strongest baseline, the reactive A*-DWA, it reduces peak risk by 22.9% and average risk by 19.0%. In the ten-AGV scenario, peak risk falls by 28.1% and conflict frequency by 36.5% relative to A*-DWA, while the fleet makespan is 5.8% shorter. (4) Conclusions: Risk-aware prediction and adaptive local planning improve the safety and the efficiency of multi-AGV navigation in dynamic warehouse environments; the cooperative layer contributes robustness and bounded waiting at high traffic density rather than throughput at the densities tested here.
No takes yet. Share an insight, caveat, or question.
Lyu et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: