Grid Inference and Partial Scan Registration for Intelligent Collaborative Robot Systems

Abstract

This thesis proposes advancement of the collaborative and intelligent abilities of Tutor Intelligence robot systems through leveraging the geometry of array structures to perform online inference of object locations and registering partial in-hand scans to automatically orient objects. This research will automate portions of the data annotation process required for the robots’ deep intelligence, enabling the collaborative robot systems to more efficiently and effectively perform pick-and-place tasks. Evaluation is conducted through an exploratory pilot study, and further design recommendations are given.