Fusion of Improved A* Algorithm and Reinforcement Learning for Path Planning of Construction Transport Robots

Path planning is crucial for ensuring the safe and efficient operation of Material Transport Robots (MTRs) in construction industrialization. To overcome the limitations of traditional methods, such as poor dynamic obstacle avoidance and the global-local decision-making mismatch, this study proposes a collaborative optimization framework that integrates an improved A* algorithm with Proximal Policy Optimization (PPO). The framework employs an enhanced A* algorithm for global planning and a PPO-based mechanism for local decision-making, fused via a hierarchical architecture. Simulation results demonstrate that the proposed method achieves a 100% success rate in dynamic environments, where all benchmark methods fail. Furthermore, it reduces planning time by approximately 78.5% and significantly improves path smoothness in static complex scenarios. This study effectively mitigates path oscillation and local minima issues, providing a robust and reliable path planning solution for intelligent construction.