Efficient Object Manipulation in Confined Space Using Compliant Based Gripper

Abstract

Abstract Manipulating small objects within a confined space is one of the challenging scenarios that industry 4.0 is currently facing. This is specifically important if the task on hand is done in human-robot collaboration. In such scenarios, the human operator may not necessarily place all the items in the most optimum locations. In such cases, it is necessary for the robotic system to identify the objects and isolate any items that are colliding. This should be followed by the priority identification for relocating the scattered objects within the given space to accommodate any incoming objects. This paper addresses the above-mentioned issue by providing a machine vision algorithm where the object added by a human collaborator are identified and then prioritized for the robotic manipulator to relocate them and provide space for any incoming objects. An optimization problem is formulated to identify the most suitable place for any of the objects that need to be replaced by the robotic manipulator. Using an adjustable stiffness gripper, we then introduce four basic actions (inner/outer grasp, rigid push, compliant push) to perform the manipulation task while avoiding any collision with the environment and other objects. The efficiency of the proposed work is demonstrated in several different cases of object manipulations.