基于模仿学习的塔吊路径规划模型研究OACHSSCD

Tower cranes is one of the most critical construction machinery due to their high lifting ca-pacity and extensive coverage,face the research challenge of achieving safe and efficient material transportation in the context of smart construction.However,traditional path planning algorithms pri-marily focus on the spatial geometric feasibility of the paths,often neglecting practical constraints such as the operating habits of tower crane operators.To enhance path planning feasibility,this study pro-posed an imitation learning(IL)-based path planning framework.Firstly,expert lifting data were col-lected through a virtual reality(VR)driving simulation system to construct a training dataset,and deep neural networks were employed for behavioral cloning(BC)to learn expert strategies.Subse-quently,the proposed method was compared with reinforcement learning approaches in cases using key performance metrics.Experimental results demonstrated a 21％reduction in path complexity,a 7％improvement in planning success rate compared to reinforcement learning solutions and RRT.The study validates the effectiveness of imitation learning in generating efficient and feasible paths for com-plex construction scenarios,offering a novel approach to address path planning challenges in intelligent tower crane operations.