Multi-agent Path Optimization Using MMD*L and SCP With Terrain Information

This paper presents a terrain-aware multi-agent path planning framework integrates multiple-mission D* lite (MMD*L) for lobal path planning and sequential convex programming (SCP) for local trajectory optimization. The proposed approach enhances mission efficiency by incorporating three key terrain factors—traversability, elevation, and visibility—to derive optimal paths in complex environments. The MMD*L algorithm efficiently assigns waypoints to agents using the multiple-traveling salesman problem (mTSP) formulation, which optimizes waypoint allocation for each vehicle. For local path planning, SCP is applied to generate real- time trajectory adjustments that respect vehicle dynamics while ensuring obstacle avoidance. Simulation results demonstrate that MMD*L significantly reduces computational overhead in dynamic environments by reusing precomputed graphs, whereas conventional A* must recompute paths entirely when the starting position changes. Furthermore, SCP-based trajectory optimization successfully generates smooth and dynamically feasible paths while effectively avoiding obstacles. The proposed framework improves real-time adaptability and computational efficiency, making it well-suited for applications in autonomous ground vehicles, unmanned aerial vehicles, and robotic swarm coordination.