Automated Layout Planning for Reconfigurable Robotic Assembly Cells

Abstract

Flexible and reconfigurable robotic systems can minimize the manual effort by providing planning algorithms to react to changing requirements. An important part of the planning process is the initial design of a robotic workcell layout. The components require to be arranged in such a way that each task is reachable by the manipulator and no collisions occur. If the execution area is limited in size, several robots are involved or the tasks are complex, planning a suitable workcell layout becomes a challenging problem. In this work, we present a framework which suggests appropriate workcell layouts by analyzing the kinematic properties of tasks and workcell components such as storages and robots. We formulate a constrained nonlinear optimization problem on discrete parameter sets to optimize the layout regarding the placement of the components. The general framework provides an interface to optimize multiple objectives. In particular, we analyze the maximization of the region in the workspace in which a task could be executed. Constraints guarantee the absence of collisions and the reachability of each task. The framework is evaluated for three scenarios, where two robots have to cooperate to insert a slot-nut into an aluminum profile. The optimization problems are solved by three optimization algorithms, namely the pure random search, hill climb and genetic algorithm, where the genetic algorithm has slightly outperformed the others. A feasible layout has been found within milliseconds. The presented approach provides a powerful framework to optimize the layout of a robotic workcell for flexible systems with multiple tasks, storages and robots.