Machine-vision-based teleoperation aid

Abstract

When teleoperating a robot using video from a remote camera, it is difficult for the operator to gauge depth and orientation from a single view. In addition, there are situations where a camera mounted for viewing by the teleoperator during a teleoperation task may not be able to see the tool tip, or the viewing angle may not be intuitive (requiring extensive training to reduce the risk of incorrect or dangerous moves by the teleoperator). A machine vision based teleoperator aid is presented which uses the operator's camera view to compute an object's pose (position and orientation), and then overlays onto the operator's screen information on the object's current and desired positions. The operator can choose to display orientation and translation information as graphics and/or text. This aid provides easily assimilated depth and relative orientation information to the teleoperator. The camera may be mounted at any known orientation relative to the tool tip. A preliminary experiment with human operators was conducted and showed that task accuracies were significantly greater with than without this aid.Keywords: Machine Vision, Teleoperation, Telerobotics, Pose Estimation. , INTRODUCTIONTelerobotics has the potential to greatly benefit many space applications, by reducing the great cost and hazards associated with manned flight operations. For example, space assembly, maintenance, and inspection tasks can potentially be done remotely using robots instead of using extra-vehicular activity (EVA). Teleoperation is an attractive method of control of such robots due to the availability and maturity of the technology.Unfortunately, using remote camera views degrades the operator's sense of perception as compared to actually having the operator physically on the scene. This paper describes how artificial intelligence (specifically, machine vision) can be used to implement a teleoperator aid that improves the operator's sense of perception. The Problem of Perception in TeleoperationIn this paper, we are concerned with the class of teleoperation tasks that involves placing the endeffector of the robot in a certain pose (position and orientation) relative to some other object in the scene. This class of tasks includes most manipulation tasks, since generally an object must be grasped or manipulated in a specific manner, and so the accurate placement of the end-effector with respect to that object is a required precondition. In addition to manipulation requirements, the end-effector must be moved accurately around the workspace to avoid collisions. We are also concerned with the class of tasks in which the identity, geometry, and appearance of the object to be manipulated is well known in advance, but its location is only approximately known. Finally, we are concerned with tasks where the end-effector must be placed relative to the object with an accuracy that is tighter or more stringent than the initial a priori knowledge of the location of that object. This situation is common when the task and environment are fa...

A Variation Reduction in the Tele-Abrasive System: A Study of Human Movement

“…Consequently, the operator must be intensively trained to ensure that he or she is capable of performing the task competently and efficiently. Among the sources of variation mentioned earlier, recent research has proposed many approaches in an effort to reduce variation by improving visual perception [9], stabilizing time delay [10], and maximizing robot accuracy [17][18][19]. However, in teleoperation tasks, although human arm movement is a main part of a shared-control system, human movement error has received very little research attention.…”

“…Despite the many benefits of a tele-abrasive system, one of its significant drawbacks is the system variation that degrades the accuracy and precision of the abrasive nozzle position. There are several sources that cause system variation, such as visual perception [9], time delay [10], robot calibration [11], and human movement error [12]. Among these factors, human movement error in a tele-abrasive task has a unique characteristic and has not been solved.…”