A path planning algorithm combining Levy flight and probabilistic roadmap methods(LPRM)is proposed. The Levy flight method is applied to narrow area sampling,random points in obstacles are walked to free space by Levy flight,and the collision test is extended to ensure that the sampling points are located in the narrow area,which improves the sampling quality and efficiency in narrow areas.To avoid the generation of invalid points,the map is pre- processed before sampling,the obstacles are inflated and their boundaries are extracted,and the number of sampling points in narrow areas is calculated based on the boundary information,ensuring a reasonable distribution of sampling across the map.Further,considering the actual working condition of the mobile robot,the path trajectory is optimized by using segmented Bessel curves to conform to its kinematic constraints and improve the mobility of the mobile robot. The simulation experiments compare three algorithms,LPRM,traditional PRM and bridge test,under different environment maps,and the results show that the LPRM algorithm improves the planning efficiency by 35.1%and 32.2%respectively compared to both in a single narrow area environment,and its planning efficiency improves by 32.9%and 15.5%respectively in a complex environment,and reaches convergence 400 and 100 sampling points earlier, the planning efficiency and success rate improved significantly,with shorter overall time consumption and better paths, which can reduce the energy consumption of the mobile robot itself and improve the overall work efficiency.