A cloud-based control architecture design for the interaction of industrial robots with soft objects

Abstract

In order to control the interaction of industrial robots with soft materials many computationally-intensive sub-tasks have to be performed in parallel, such as material simulations, control policy optimization or path planning among many others. Since the robot control cannot process the amount of information needed for those tasks, the whole system needs to be split up in two parts: the physical system and the computationally demanding software that runs exclusively in a cloud environment. We analyse the requirements, make the assumptions and define the scope of the system before proposing a novel cloud-based control architecture for industrial robots to manipulate soft materials. A system is set up and evaluated with future challenges identified for the further specification and implementation of the proposed control architecture.