KDI Home    Contact Us   

Collaborating in Shared Visual Environments

Shared visual environments substantially enhance the performance of collaborative tasks, according to researchers at Carnegie Mellon University (CMU). Their research has significant implications for practical uses such as telemedicine, distance learning, and the staffing of help desks for repair.

The work has been carried out with support from the National Science Foundation's KDI program.

When people work together performing a task in real time, they attempt to maintain an up-to-date mental model of what the others are likely to know and do at any moment. They use this model to shape what they say and how they collaborate. The research at CMU's Human-Computer Interaction Institute examined how speakers and listeners used shared visual spaces—a joint view of a physical area—to form this mental model when carrying out collaborative physical tasks. These were tasks in which two or more individuals worked together on objects —for example, by repairing a bicycle, building a large toy robot, or completing a jigsaw puzzle.

In their experiments, the researchers manipulated the extent to which partners shared visual space by the technology provided them during the task—face-to-face settings versus other modalities, and displays with or without delays.

Robert Kraut, Herbert A. Simon Professor of Human-Computer Interaction at Carnegie Mellon, says, "Across our many studies, we have found consistent evidence that participants in these tasks make use of shared visual space to infer one another's knowledge and intentions." They found that the availability of shared visual space enables helpers to anticipate their partners' actions and provide assistance in a timely fashion.

The researchers experimented with adding to the visual displays technology to show the focus of attention of the participants' eyes. However, according to Kraut, such eye-tracking technology was not found to improve communication back and forth. This is most likely because the field of view provided by the camera in which the eye-tracker was situated was too narrow; in other words, it did not create enough shared visual space between partners. In addition, it may be hard for remote helpers to follow the video feed when cameras are head-mounted and thus move around continuously.

On the other hand, according to Kraut there was a considerable improvement in communication when participants used display wands such as those that TV football commentators use to diagram plays on the video screen.

Another member of the Carnegie Mellon research team, Research Scientist Susan Fussell, notes, "The research paradigm that we work on is a remote instruction paradigm, where one person is telling another person how to do a task with physical objects. I think all the results apply to education and to help desks, scenarios like that. All of the results of our work are that it would be very beneficial to have a shared visual space."

"We've come up with various inexpensive ways to create this space," Fussell adds. "For example, right now we have a network camera that has its own IP address, and it focuses on the work space. You can access it over the Internet using any sort of browser, so you can see what's going on from anywhere with a very inexpensive tool. We think that that would have a lot of applications in education."

The research team working on the Carnegie Mellon project has been multi-disciplinary, involving computer scientists, social psychologists, and cognitive psychologists.

Fussell comments, "This particular grant I think was very beneficial for students. … There were a lot of graduate students and very large number of undergraduates working for us, and they went on to master's and Ph.D. programs. The training aspect of it was excellent." She estimates that a total of 25 to 30 students were involved over four years.

Back to Top of Page