David Drascic's Research

Last Update: Fri Jun 20 06:36:07 EDT 1997

My work began by focusing on improving the human-machine interface for bomb-disposal telerobotics. It subsequently followed two paths: improving the display for telerobotics by using stereoscopic video, and improving the input mechanism through application of Augmented Reality.

My early work at the ETC-Lab focussed on the costs and benefits of using stereoscopic displays for teleoperation. As a quick summary of my findings, I would say:

For some tasks, like teleoperation, stereoscopic displays are a great idea, being easier to learn than monoscopic displays, and enabling faster performance with fewer errors. [HFS Conference 1991]
For other tasks they may not be worth it. It really depends on the task and on the display being considered. [Master's Thesis, DND Workshop 1993]
Tasks that are highly repetitious won't benefit as much from stereo displays as will those that are always unique, such as bomb-disposal and hazardous materials cleanup.
Displays in which the camera views are restricted to a single position, and those in which a significant proportion of the motion of the system is perpendicular to the display will benefit most greatly from stereoscopic displays.
Tasks that involve only computer graphics without any stereoscopic video can often be accomplished better by redesigning the task than by using stereoscopic displays. Stereoscopic perception works best in fully rendered and richly textured environments like the real world. Few computer-based displays can afford the desired degree of realism.

Augmented Reality refers to displays that combine real world images (either video and/or direct view) with computer-generated images. Unlike Virtual Reality, which seeks to create an entirely artificial world, the goal of Augmented Reality is to present to the user the real world (possibly at a remote location, such as underwater or space, or of a different scale, such as micro-surgery) that is enhanced (or "augmented") with computer graphics. The applications of this technology include tele-operation (using a robot to do a job at a distance using remote control), medical and architectural imaging, and many other areas.

There are several different approaches to achieving Augmented Reality. At the University of Toronto, we combine stereoscopic video with carefully calibrated stereoscopic computer graphics. At the moment, I am investigating the perceptual aspects of Augmented Reality displays. In particular, I am examining the effects of accommodation-vergence conflict and accommodation mismatch on the perception of depth in stereoscopic displays.

Working with Prof. Paul Milgram and Dr. Julius Grodski, I pioneered the ETC-Lab's work in Augmented Reality, culminating in our patented technology which we call ARGOS (Augmented Reality through Graphic Overlays on Stereo-video). One of the first useful examples of Augmented Reality was the virtual pointer, which consisted of a carefully calibrated stereoscopic graphic pointer superimposed on a painstakingly designed stereoscopic video system. The pointer can be moved freely around in the remote view, and used to measure distances and locations of objects in the real world, and can pass this information on to the computers controlling the robots. [SPIE Stereoscopic Displays 1991, IROS 1993]

Most of my effort since developing this technology has been devoted to the usability issues of stereoscopic displays, and the perceptual aspects of Augmented Reality displays. At the moment, I am trying to determine if stereoscopic displays that combine computer generated objects with objects presented via stereoscopic video and objects viewed directly will work. Will people see things where the designers hope they will? Or will distortions, depth cue conflicts, and unforeseen perceptual issues introduce uncorrectable errors?