A path planning method based on improved RRT (rapidly exploring random tree) algorithm is proposed for the motion obstacle avoidance problem of a composite mobile robot with a 6-DOF mechanical arm in guide roller production line. First, the kinematic model of the 6-DOF manipulator is established by using the MD–H (Modified Denavit–Hartenberg) parameter method, revealing the mapping relationship between the joint space and the operation space of the manipulator. Second, a trajectory planning algorithm based on quintic polynomial interpolation and a path planning method based on improved RRT algorithm are proposed to carry out motion planning for the manipulator. Finally, the above methods are simulated in MATLAB, and the simulation experiment shows that the motion trajectory of the proposed method can ensure the smooth operation of the robot, and the continuity is improved. There is no obvious vibration during the entire process from the starting point to the end point, and the manipulator motion is stable. The method is feasible. The above characteristics prove the feasibility of the proposed methods. Simulation of the production workshop environment for guiding roller in ROS (Robot Operating System) was conducted. The loading and unloading tasks of steel shaft parts for the guiding roller were completed. The specific application of the proposed motion planning method in the simulation environment was realized, demonstrating the feasibility and effectiveness of the method.