Online time-optimal path and trajectory planning for robotic multipoint assembly

Purpose The purpose of this paper is to propose an efficient path and trajectory planning method to solve online robotic multipoint assembly. Design/methodology/approach A path planning algorithm called policy memorized adaptive dynamic programming (PM-ADP) combines with a trajectory planning algorithm called adaptive elite genetic algorithm (AEGA) for online time-optimal path and trajectory planning. Findings Experimental results and comparative study show that the PM-ADP is more efficient and accurate than traditional algorithms in a smaller assembly task. Under the shortest assembly path, AEGA is used to plan the time-optimal trajectories of the robot and be more efficient than GA. Practical implications The proposed method builds a new online and efficient path planning arithmetic to cope with the uncertain and dynamic nature of the multipoint assembly path in the Cartesian space. Moreover, the optimized trajectories of the joints can make the movement of the robot continuously and efficiently. Originality/value The proposed method is a combination of time-optimal path planning with trajectory planning. The traveling salesman problem model of assembly path is established to transfer the assembly process into a Markov decision process (MDP). A new dynamic programming (DP) algorithm, termed PM-ADP, which combines the memorized policy and adaptivity, is developed to optimize the shortest assembly path. GA is improved, termed AEGA, which is used for online time-optimal trajectory planning in joints space.

Automated summary

This paper proposes an efficient path and trajectory planning method for online robotic multipoint assembly. The combination of PM-ADP and AEGA algorithms is shown to be more efficient and accurate in smaller assembly tasks, optimizing both path planning and trajectory planning.