With the rapid development of Autonomous Ground Vehicles (AGVs) technology, the path planning capability of AGVs in unknown and complex environments has become a research hotspot. Focusing on the path planning task of differential drive vehicles in complex environments without an initial map, we propose a path planning algorithm framework that integrates the vehicle’s dynamics model. This framework aims to address the limitations of traditional path planning methods, which overly rely on global maps and neglect the dynamics characteristics of the AGV. In this paper, we adopt the Kinodynamic A* algorithm and design a corresponding heuristic function. By establishing a temporary map based on real-time radar information and integrating the dynamics model of the differential drive vehicle, we plan a collision-free shortest path that ensures the stability of the vehicle during high-speed motion. Additionally, the path is re-planned in real-time as the vehicle tracks the planned route towards the target point. This approach reduces the difficulty of reaching the target point without an initial map and enhances the robustness of the planned path. It provides an effective solution for path planning of AGVs in dynamic environments.