Abstract: A path smoothing optimization algorithm integrating the D*lite algorithm and the artificial potential field (APF) method was proposed to address issues such as long search time, unsmooth paths, and insufficient obstacle avoidance performance in path planning for robots in dynamic environments. The heuristic function and key value operator of the D*lite algorithm were optimized to improve search efficiency. The repulsive force function of the APF was improved and random virtual obstacles were introduced to resolve the problem of robots easily falling into local optima, enabling target approach and dynamic obstacle avoidance through an attractive-repulsive force coordination mechanism. The two algorithms were integrated to achieve rapid path search and efficient obstacle avoidance in dynamic environments. To further enhance the practicality and motion smoothness of path planning, a pruning strategy was adopted to remove redundant nodes, reduce computational complexity, and decrease unnecessary turns to lower path tortuosity. On this basis, a cubic B-spline curve was introduced to smooth the path, eliminate sharp turning points, and generate a smooth path that conforms to kinematic laws. Simulation results demonstrate that compared with the traditional D*lite algorithm, the proposed algorithm reduces the search time by 13.3%, 9.3%, and 11.5% in three typical simulation environments—simple, complex, and narrow passage respectively, significantly improves path smoothness, and effectively enhances obstacle avoidance capability. Ultimately, an organic integration of efficient global planning and local dynamic obstacle avoidance is achieved, the path search time is significantly shortened, and the superiority and practicality of the proposed algorithm are verified.