Ninth Annual Conference
CHI '91 - Reaching Through Technology
by Ben Delaney

This article originally appeared in CyberEdge Journal #3, May/June © 1991

CHI '91, the annual conference of the ACM Special Interest Group for Computer/Human Interface (SIGCHI) was held on April 28th through May 2nd in New Orleans. CHI '91 brought together over 2,100 representatives of virtually every important company and university doing interface research anywhere in the world. Occupying four floors of the Sheraton, it was a dawn to dusk, nonstop cacophony of presentations, demos, meetings, chance encounters, practical jokes and general enthusiasm. It was also a most important conference for those interested in improving the way computers and people interact.

Conference co-chairs Keith Butler of Boeing and John Thomas of NYNEX and a multitude of volunteers did a great job of assembling a full schedule of exciting presentations. Our only complaint was that too many interesting presentations were scheduled simultaneously, preventing us from attending all of them. The conference went off like clockwork, with presentations starting on time, the AV and computer equipment working, and adequate opportunities to meet and mix with colleagues.

The CHI '91 program was ambitious and rewarding for those with plenty of energy. Included on the program were 30 Tutorials, an intensive Doctoral Consortium, eight Workshops, about 20 SIG meetings, more than 20 Technical Presentations, formal and informal Demos, formal and informal Video Presentations, many Short Talks, more than half a dozen Panel Discussions, dozens of Posters (showing work too recent to include on the formal program), Lab Overviews, Vendor Exhibits, and the Interactive Experience.

Hands on

The Interactive Experience was a demo area where conferees could experience twelve different examples of innovative interface technology in a non-structured regime. Exhibits ranged from advanced hypermedia, through a novel keyboard concept, to 3D sound and full-immersion virtual reality. Open long hours, the Interactive Experience allowed unrestricted interaction with the developers and the systems, for learning and experimentation.

The Interactive Experience featured four systems which construct Virtual (first person, immersive) or Artificial, (third person, observing) Realities. The artificial reality demos included Vivid Effect's Mandala System, which enables the participant to interact with programmed environments. The participant saw herself in a monitor below the video camera capturing her image. The live and stored images are combined in the computer.

David Rokeby demonstrated A Very Nervous System, similar to the Mandala System in that it used video input, but unique in mapping the participants into an audio space where their movements played instruments and triggered other sounds.

The two Virtual Reality exhibits included IBM's demo of Rubber Rocks, a two participant system using gloves and head tracking, and large monitors. It depicts objects with variable characteristics of mass, hardness and breakability. Participants were indicated as hands in the scene, and played various games with the objects. Dan Ling, of the Veridical User Environment Group explained that Rubber Rocks is an example of real-time control of a multi-user, virtual environment. It runs on heavy-duty hardware: seven RS6000 systems plus various other ATs and dedicated systems.

Rubber Rocks demonstrated concurrent, high performance computing with synchronized multiple events, including simple, trainable voice recognition, vocal feedback, 3D graphics generation, two user interaction, and sound generation.

Across the aisle from IBM was the other VR exhibit, Boeing's Chris Esposito giving tours of a virtual aircraft, the VSX. Using a VPL RB2 system and a model of the VS-20 tilt-rotor aircraft, Boeing has developed a conceptual model of Virtual Reality as a design tool. Participants donned goggles and glove and moved around and through the aircraft. It was possible to fly the aircraft and rearrange the cabin. Boeing had presented the VSX at the Industrial Symposium on Virtual Worlds in February; the current version is running at roughly twice the frame rate which facilitates a much smoother, more comfortable interaction.

In the exhibit hall, sixteen commercial exhibitors presented their wares. Half of the vendors were publishers, showing a large number of titles related to interface design and related topics. The rest showed software, testing tools and various products of interest.

The presentations

As much of the material presented at CHI '91 was concerned with traditional human factors questions; i.e.; where do you put this control, how do we eliminate operator error. We have concentrated on VR and advanced technology presentations. Here are highlights of some of the most interesting presentations at CHI '91. Full proceedings are available from ACM, at the address listed at the end of this article.

The opening plenary speaker was Thomas Allen, Ph.D., Director of MIT's International Center for Research on the Management of Technology. He spoke on the difficulties of managing basic technology research. He focused on the barriers to cooperation and the acceptance of change within an organization, and offered his experience in overcoming these barriers. He noted that different corporate cultures and structures contribute to the problems in technology transfer that often arise. Rewarding the "gatekeepers" in an organization, the people who are high-performing, outward-looking experts, will significantly abet the necessary cross-communication between scientists and technologists. He stressed the need for understanding people and the techniques to bring them together, since that will improve and encourage collaboration in both collegiate and corporate environments.

Meanwhile, back in the lab

In a Lab Overview, John Richards of IBM's Watson Research Center described their work in Usability Engineering, Multimedia Applications and User Interface Theory and Design. This all started, he explained, when in 1985 "IBM realized that usability really did matter." The work includes defining usability issues, standardizing tools for interface development, and extensive testing of theory against user expectations.
In another Lab Overview, Ron Becker, Marilyn Mantei, William Buxton and Eugene Fiume of the University of Toronto described, among other efforts, the group's work in "virtual offices" and computer support of collaborative work. Their work builds on Buxton's work at Xerox EuroPARC. EuroPARC's related project was discussed by Michael Travers at the Second International Conference on Cyberspace (see page 1) and at CHI. The University of Toronto group is also exploring the perceptive and psychological aspects of collaborative work, and interfaces in general.
Information ala carte

In a Technical Session, Stuart Card, George G. Robertson and Jock Mackinley of Xerox PARC presented the Information Visualizer, a windowed, animated, integrated system for information management. It consists of four major modalities for the representation of different types of data; Hierarchical, Structural, Continuous and Linear. These all share a mouse-controlled, 3D windowed interface, using a metaphor of interconnected (3D) rooms, as opposed to (2D) windows. Each data model incorporates an animated 3D structure. Animation is used to maintain the spatial relationships of data elements while an entire structure of which they are part rotates or shifts in relation to the viewer. Part of the success of this model stems from the PARC team's understanding of optimal response times and animation speeds, which they incorporated into the Information Visualizer. The motion feels right, and it is easy to track a piece of data contained in a rotating hierarchical structure.

The VR crowd

Wednesday morning dawned bright, as the rain and tornado warnings of the two previous days had abated. The day started with the one session in the conference specifically targeted at Virtual Reality. This session drew a standing room only crowd, estimate at 650 delegates. Presentations were made by an IBM team from the Watson Research Center, Randy Pausch of the University of Virginia, a group from Boeing and a team from the university of New Brunswick.

J. Bryan Lewis, Lawrence Koved and Dan Ling of IBM discussed their requirements for acceptable performance in virtual systems: real-world performance, simultaneous use of multiple, concurrent input and output devices, good project coordination to produce consistent, interactive modules, and great flexibility. They explained how they have devised rule sets that facilitate inter-process communication in virtual world systems. The process they described was demonstrated in the Rubber Rocks demo.

The next speaker was Randy Pausch, who excited the audience with his explanation of a home-brew VR-based system which cost only $5.00 per day. Lacking an adequate budget to purchase a VR system, Pausch built his own. He combined two mechanically linked Private Eye displays, a Mattel PowerGlove, and one Polhemus 3Space tracker. The system provides 720 by 280 spatial resolution and displays wire-frame graphics generated by a 80386-based, 2.5 MIP, PC clone system. Including the voice input which he intends to add, Pausch calculates the total system cost at under $5000, which when amortized over the typical three year life of the equipment, equals a cost of about $4.55 per day. He is now soliciting support to build 10-20 such systems, providing access to VR to an entire graduate class.

Chris Esposito of Boeing gave the next presentation, in which he discussed the VSX project shown in the Interactive Experience. The VSX was developed with Meridith Bricken of the Human Interface Technology Lab, and Keith Butler, also of Boeing. As previously described, it is a virtual reality model of an aircraft. Esposito discussed some of the project's constraints and processes, and explained how Boeing hopes to use VR constructs to aid in design of future craft. Chris was a busy man at this conference; in addition to his presentation, he was conducting demos and had responsibility for all computer support for the entire conference. He did a great job, and confided that a week of relaxation in Hawaii was his next big project.

The final VR presentation was made by Dr. Colin Ware of the University of New Brunswick. He and Leonard Slipp have been working on quantifying the relative performance of various VR interface devices, specifically the Spaceball, Flying Mouse (which they call a Bat), and a standard, 2D mouse with virtual control panel. By designing a three dimensional "horn of plenty" shaped maze and measuring the time required to navigate it with different devices, they were able to compare the effectiveness of each controller. They discovered that the mouse and control panel was by far the least acceptable control device, while the Flying Mouse was the easiest to use.

And Tom Furness

The closing plenary featured Dr. Thomas Furness, Director of the Human Interface Technology Lab (HIT Lab) at the University of Washington. Dr. Furness spoke on the emergence of virtuality as a technique of computer interfacing. He discussed his agenda at the HIT Lab and the staff he has assembled to accomplish those goals. He described the accomplishments of his Human Engineering Division at Wright Patterson Air Force Base, where he was instrumental in development of the Super Cockpit, a virtual cockpit for advanced fighter planes. He concluded with a series of observations: "I've learned the importance of seeing. It's important to be able to see into new worlds, to make the next jump in civilization. We need to solve real problems. We need to know the user, to know what's in the user's head. We need to build stuff. We spend a lot of time speculating", he said, "but nothing takes the place of actually building something, no matter how crude. We should anticipate a VR winter, as we saw an AI winter", and finally, "VR is not an end, but is just another tool." One use of that tool is education. He noted that the money paid to the top three NFL draft choices could end the educational salary crisis in the state of Washington. Declaring that no investment is more important than educating our children, he cautioned, "I really believe we need to get our priorities straight". He asked the audience to help make that happen. They responded enthusiastically.

Why is CHI important?

First and foremost, it is one of the largest interdisciplinary meetings related to computing. Much of its value lies in the opportunity to discuss interface problems with experts in disparate fields. You can always find somebody to discuss your pet project or offer solutions to your current stumbling block. The Interactive Experience offers a chance to actually get your hands on some equipment and see what works and what doesn't. This alone is worth the price of admission for many. We recommend CHI to those who are involved in VR applications or other advanced interfaces. You are sure to learn something, and have a great time doing it.

Full Conference Proceedings are available from ACM. Next year, CHI '92 will be held in Monterey, California on May 3-7, 1992.

Contact: SIGCHI, c/o ACM, 11 w. 42nd Street, New York, NY, 10036 USA, 212 869-7440.

Questions? More Info? Email the Webmaster

407 M. L. King Jr. Way, Oakland, CA 94607 USA
+1 510 419-0800

Our privacy policy may be read here.