Implementation of near-time-optimal inspection task sequence planning for industrial robot arms

Abstract

Near-time-optimal inspection task sequence planning using industrial robots is implemented. In the system, one robot arm holds an inspection tool while the part to be inspected is held by a second robot arm. A time-efficient continuous joint acceleration profile is proposed and simplified by suitably choosing the ratio of time intervals of the profile. Travelling times between any points are calculated based on limits of joint positions, velocities, accelerations and jerks. Time-efficient joint trajectories are generated between all pairs of points. Then the inspection task sequence planning is formulated as a variation on the travelling salesmen problem and simulated annealing is introduced to find a near-time-optimal route. A hierarchical structure is presented which integrates task planning, time-efficient trajectory planning, control of two RTX SCARA type industrial robots, and graphic display of the route. A SUN SPARC-5 workstation and a transputer network are used. Process 1 of the workstation is used for task planning and graphic display. Process 2 of the workstation interfaces to a transputer network used for real-time time-efficient trajectory generation and control of the two RTX robot arms. Experiments have been made on the system based on the theoretical computational results. The experimental results are satisfactory and consistent with their theoretical counterparts. The effectiveness, efficiency and feasibility of this implementation have thus been verified.