Abstract: Aiming at the problem that ordinary wheeled robots are difficult to pass through a narrow working area indoors, and are easily trapped in a local area and unable to complete navigation, a local path planning fusion algorithm for Mecanum wheeled robot chassis was proposed. Based on the advantages of the Mecanum wheel, the Dijkstra global path planning algorithm was used to plan the global path, and the Dijkstra and PID algorithms were combined to control the local path planning. According to the attitude information, the robot speed was decomposed laterally and vertically, the rotation speed was limited to drive the robot to achieve stable movement while keeping the attitude unchanged. The Mecanum wheeled robot equipped with the robot operating system was used for navigation test to verify the effectiveness of the fusion algorithm in this paper. The results show that in a complex and narrow indoor environment, the algorithm can achieve autonomous navigation and obstacle avoidance of the robot, and can successfully reach the destination from the set starting point. Compared with the traditional time-elastic band (TEB) algorithm, the navigation time of this algorithm in an obstacle-free environment is reduced by 30.56%, and the navigation success rate in an obstacle environment is 4% higher, which can meet the demand of indoor mobile robots to reach the navigation target point quickly.