Simulation of Grasp Localization Inferences for a Dual-Arm Collaborative Robot

Abstract

Abstract Grasp Localization is an important problem in robotic manipulators which fundamentally aims at estimating a pose for the end effector which ensures that the object intended to be grasped does not slip off. This estimation is achieved through exteroceptive sensors used in robots such as vision systems which through features and learning-based approaches. One of the fundamental means to validate such an approach is by simulating the grasp routine for any given geometry and its probable grasp poses. This paper presents the details of a simulation work involving a dual-arm collaborative robot with each arm consisting of seven degrees of freedom. The dual-arm robot performs a grasp routine for a geometry that is imported from CAD software. To perform the simulation the various attributes of the robot are to be observed since the simulated robot is a mimic of a commercially available one.