Robust online multi-robot path planning for complex environments

This work develops an adaptive enhanced framework to address slow convergence to high-quality trajectories, insufficient path smoothness, and elevated collision risk in multi-robot path planning. The approach is embedded in a multi-constraint, single-objective optimization model that incorporates kinematic limits, static and dynamic obstacle avoidance, and practical mission cost. Three complementary mechanisms are introduced: a velocity-alignment coordination to accelerate cooperative convergence of candidate solutions; an adaptive follow escape probability strategy that dynamically balances global exploration and local refinement to produce smoother and safer paths; and an adaptive random-perturbation mechanism to escape local optima and increase robustness in complex scenarios. Benchmark and simulation studies show that the proposed framework outperforms baseline and state-of-the-art comparison methods in terms of convergence accuracy, convergence speed, computational efficiency, path quality, and safety, providing a practical and robust solution for improving autonomy and operational performance in complex engineering multi-robot systems.