To address the limitations of traditional path planning algorithms in autonomous driving emergency scenarios, such as environmental modeling distortion, insufficient path search efficiency and safety deficiencies, this paper proposes a high-precision urban electronic map-based “global-local” coupled path planning framework. By integrating an improved D*Lite algorithm with the dynamic window approach (DWA), the framework achieves coordinated control of dynamic global path optimization and local obstacle avoidance. In global planning, a five-neighborhood search strategy replaces conventional eight-neighborhood search to prevent path tortuosity, while a multi-objective cost function incorporating risk coefficients significantly reduces cumulative path risks. For local planning, a risk-aware dynamic evaluation function is designed to enhance the real-time performance and safety of obstacle avoidance. Simulation results demonstrate that compared with existing methods, the proposed coupled algorithm achieves significant improvements in path planning efficiency, path safety, and trajectory smoothness. Further verification through typical emergency scenarios including traffic collision avoidance and sudden passenger demand responses confirms the algorithm's robustness, providing theoretical support for safe navigation of autonomous vehicles.