Parallelism of Pick-and-Place operations by multi-gripper robotic arms

This article defines the Parallel Pick-and-Place (PPNP) problem and develops a framework for optimization of its operations performed by multi-gripper robotic arms. The motivation lies in the lack of analytical methods for the parallelism of structured/unstructured Pick-and-Place (PNP) operations by robotic arms. Although the PPNP operations are mostly attributed to printed circuit board assembly, their applications span various other processes such as palletizing, packaging, warehousing, sorting, loading/unloading of machines, machine tending, inspection, remote maintenance, and robotic nurse assistance. Parallelism of the PNP operation is enabled by facilitating the robot's end effector with multiple grippers and magazines in order to perform simultaneous pickups and placements of items. Two different formulations of the PPNP process are developed regarding two cases: (1) Optimal routing while the pickup and placement positions are fixed; (2) Optimal routing and configuration of pickup and placement positions at the same time. An efficient swarm intelligence algorithm based on Ant System and Tabu Search is developed for handling the complexity of the PPNP problem. Through a reinforcement learning mechanism, the robot is provided with a certain level of intelligence to adapt to changes in its working environment and find the shortest route automatically, after relatively few computational iterations. Results of several experiments indicate superiority of the developed framework for the PPNP operation to conventional approaches in terms of cycle time, as an indicator of the overall movement distance and energy consumption. Published by Elsevier Ltd.