Abstract: For a problem that rescuing robot was easily stacked into local minimal point by using traditional artificial potential field method to plan path in coal mine underground with complex obstacles, a path planning method of rescuing robot based on improved artificial potential field method was proposed. In the method, perturbation field is added into gravitation field to change gravitation field function, so as to make rescuing robot go out of local minimum point autonomously when the robot stacks into the local minimum point. Meanwhile, obstacle is filled virtually through an obstacle filling method, thus a new obstacle is formed and a corresponding repulsive potential field function is generated to avoid rescuing robot to stack into local minimal point again. The simulation and test results verify feasibility and validity of the method.