Energy-Constrained Hybrid Repair for Lifelong Multi-Agent Path Finding in Smart Warehouses

Smart warehouses rely on fleets of autonomous mobile robots that must continually assign tasks, plan paths, avoid collisions, and maintain battery energy. Existing lifelong multi-agent path finding studies often emphasize travel cost or makespan, while practical deployments also involve charging, payload-dependent energy use, turning and waiting costs, and congestion. This paper presents an energy-constrained hybrid repair framework for lifelong multi-agent path finding in warehouses. The method combines a risk-aware graph representation, search-based safety repair, and learning-compatible policy modules, and it is evaluated in a reproducible Python simulator. We compare independent and prioritized A* planning, windowed cooperative planning, large-neighborhood repair, lazy configuration search, bounded conflict-based search, the proposed repair variant, and a graph neural learning baseline. The virtual evaluation reports raw task completion separately from energy-feasible completion, together with collision, charger-conflict, energy, scalability, ablation, sensitivity, and case-study measures. On 40×40 warehouse maps with 20 robots, large-neighborhood repair improves raw success from 0.345 to 0.468 relative to windowed cooperative planning and reduces energy per completed task from 369.34 to 276.83. The proposed repair variant reduces several conflict measures, but throughput remains problem-dependent. The results support energy-aware repair as a practical direction for warehouse robot coordination.