Development of sensors and algorithms for automating robotic grasping

Abstract

The difficulty of locating and grasping a target via a robot functioning as a remote teleoperator varies with the operator's perspective of the target area and the skill of the operator. To limit the amount of time and skill required to grasp a target, several methods were investigated to automate this process. These methods involved the development of sensor systems and control algorithms directed at the detection of the target in a pre-defined work area, location of the target's edges, approaching the target along its centerline, and ultimately grasping the object.\nThe specific application involved in this research concerned a mobile robot locating and grasping random objects within a room. This thesis describes the results obtained with both the tactile and the non-tactile sensors. The non-tactile sensors were chosen for the final configuration because they could locate the target more quickly and they were safer.