The Virtual Reality Roving Vehicle Project
DR. WILLIAM WINN, Director Human Interface Technology Laboratory's Learning Center University of Washington Seattle, Wash. Between November 1994 and June 1995, almost 3,000 students in grades four through 12 throughout the state of Washington experienced Virtual Reality (VR) in their classrooms. Another 365 built their own Virtual Worlds. The Virtual Reality Roving Vehicle project (VRRV, pronounced "verve") was funded by the US WEST Foundation to bring VR equipment and experiences to children in all areas of the state. The project was based at the Human Interface Technology Laboratory (HITL), part of the Washington Technology Center on the University of Washington campus in Seattle. The VRRV team consisted of software designers and engineers from HITL, Educational Technology faculty and students from the College of Education, a cadre of "Van Techs" who did everything from loading and driving vans to giving presentations and demonstrations in schools, plus students and teachers in close to 70 schools. Here, I describe what we did, how we did it, what we learned and what we plan to do next. What We Did Our goals were ambitious. First, we wanted as many children as possible to experience immersive VR and to discuss with them its potential and limitations. Popular media has hyped VR to excess and we felt that youngsters need to understand that VR is a technology that still has to mature before it is used widely. Second, we wanted to get our high-end workstations and software out of the lab and into children's hands, confident that they would use them effectively and imaginatively. And third, we wanted to see whether having students build virtual worlds that embodied concepts and principles they were learning as part of their regular curriculum would help them understand what they were studying. We were especially interested in determining whether this radically new technology and the learning strategies it supports can help children who are judged by traditional criteria to be less likely to succeed in school. Experience has shown that the only way technology can be successful in schools is if teachers and students have access to the machines they need, have the software to make them work, and have the knowledge and skill to take control of both. We addressed the access issue by setting up a van-based outreach program that carried VR workstations to schools. In the schools where worlds were built, we addressed software and control issues by providing the schools with modeling software that teachers and students learned to use, by building our own Supercard tool for automatically generating computer code that made virtual objects behave in particular ways, by putting on training workshops for teachers, and by working long hours with children as they planned and built their worlds. We used two vans. Each was equipped with a Division, Inc. Provision 100 workstation, a head-mounted display that provides stereo vision and audio, a hand-held "wand" for moving through virtual worlds and manipulating virtual objects, and an electromagnetic tracker that tracked both the position and attitude of the student's hand and head. The Provision 100 is a 486 system running UNIX. It achieves its high rate for rendering graphic images by graphics boards that provide massively parallel graphic processing. The system has sound capability. It runs dVISE software from Division (based in the U.K., with a U.S. office) that uses a C-like scripting language that serves to build and run virtual worlds. A limited amount of authoring and editing can be done through a graphic interface that runs under X-Windows. Centered Around Two Projects Our plan centered around two projects, which we called the "Hors d''euvre" and the "Entreé." All schools enjoyed the hors d''euvre; 14 stayed for the entreé. For the hors d''euvre, we took a van to each of the VRRV schools for a day, longer in a few cases. We began by giving a presentation about VR to one or two classes of students and their teachers, spending time discussing VR with them. The rest of the day we demonstrated commercially produced virtual worlds to as many as 50 students. Each student spent 5-10 minutes in a virtual world. We showed each one how to move around in the world and to interact with its objects using a "wand," a joystick-like device you hold in your hand. As you move it vertically, horizontally, towards you or away from you, the virtual hand you see in the helmet moves correspondingly. You can also tilt your hand around any axis. Five buttons control how you pick up and move objects and how you "fly" through the world. One may also walk around in the world, but this is limited by the tracker's range and the length of the cable attaching the wand and helmet to the computer. Once briefed, students put on the helmet and visited the world. For the entreé, we spent a number of days over an average of six weeks in each school working with teachers and their students. Our first visit was set up to take place at the point in the class where the teacher had taught enough content for students to be able to design effective worlds. We began our work in the schools by explaining the four-step world-building process: Planning: Students decided how they would represent the concepts and principles that their world embodied. Modeling: They drew and built their objects on Macs using Macromedia's Macromodel 3D graphics software. Programming: Students determined how objects were to interact with each other and with the participant, which guided HITL's programmers on how to put the worlds together. Experiencing, where students visited the worlds they had created and performed their assigned tasks. In subsequent visits, we worked with them at each step in the process, spending at least three days in each school. Students visited their worlds on the final visit. How We Did It We identified our initial group of VRRV schools through a questionnaire distributed at the NCCE Conference in Spokane in the spring of 1994. During the summer and early fall of 1994, teachers who had indicated an interest were invited to HITL in groups of two or three. We briefed them on the project and had them experience VR for themselves. (Other schools joined along the way.) In late fall of 1994, we held a day-long workshop for VRRV teachers and administrators on the University of Washington campus. In the morning, HITL staff gave presentations about previous work on educational applications of VR, other HITL projects and a more detailed overview of the VRRV program. In the afternoon, teachers brainstormed on how to use VR in their curriculum while the administrators met with VRRV staff to discuss how the project might best be implemented in their schools and districts. At day's end, teachers interested in being part of the world-building phase of the project were invited to submit proposals to us describing what they would like to do. Fourteen of these were selected by the end of the year; others went on a waiting list for the 95-96 school year. Worlds proposed and actually built included: A space station in which students could recycle waste materials; a medieval castle; a microcomputer, parts of which students had to assemble themselves; a rain forest whose over-exploitation resulted in an ecological disaster; a fly-over of Washington state that helped students learn their local geography; and nine others. In early 95, two full-day workshops were held for teachers selected for the entreé, during which we introduced them to the Macromodel software. Meeting in a school computer lab, the entire day was devoted to "hands-on" activities. At the end, one teacher from each school was given a set of disks to load onto their own machines. (A number of schools had to upgrade machines to 12MB of RAM and to add a math processor to run the software well.) We provided the software for free under a site license that we had purchased from Macromedia. Subsequently, several elementary teachers told us the software was too complex for their students to learn in the available time, so we modified the project for the elementary schools. Elementary students would all work on a common "Tree World" in which their task was to make a tree healthy by providing it with sunlight, water and nutrients. HITL staff built the basic Tree World. Students contributed objects that depend on healthy trees for their own survival -- everything from fruit to giraffes. We added these to the basic Tree World, creating a different version for each elementary school. In this way, everyone was able to contribute to the world without having to master 3D CAD. Finally, in February 95, we held another full-day workshop in which we covered four topics: instructional design, what VR can do best, assessment of learning from VR, and project management. At its end, each school signed up for a start date when VRRV staff would make their first visit to the school. We intended the projects to be staggered to spread our workload over the rest of the school year. Perhaps predictably, this did not work! A lot of the labor piled up at the school year's end. We also provided each school with a Teacher's Manual on world-building that we had written. The hors d''euvre had started in November and was well underway by this time. Indeed, the hors d''euvre operated fairly independently from the entreé, with its own van, personnel and computer system. We hired a pool of people who loaded, drove and unloaded the van; gave presentations to students; and ran the school demonstrations. Initially, these operated an easy day's drive from Seattle, returning to HITL at the end. Later in the year, we took vans to other areas of the state with a crew (always two people) that visited four schools a week. A lot of effort went into coordinating the program with schools throughout the state and we remain grateful to those who were flexible enough to accommodate our last-minute schedule changes caused by computer crashes, bad weather and the occasional scheduling conflict. What We Learned Needless to say, we were anxious to find out whether we had been successful in achieving our goals of helping children understand VR and of using world-building to teach content. However, our research agenda extended beyond simply determining if the students had learned anything. Any technology tends to be better at supporting some pedagogies over others. Our approach to teaching and learning was based on the premise that the acts of planning, designing and building virtual environments would help students construct an understanding of the topic they were working on. This "constructivist" approach required that most of the decision-making be placed in the hands of the students. We therefore spent a lot of time facilitating discussion in small groups of students as they argued and debated about the appearance and functionality of their world. Our role was largely one of keeping students' ambitions within the limits of what our VR systems could actually do. We were anxious to find out whether using constructivist pedagogy that VR supports so well would lead to different outcomes when compared to outcomes from students learning similar content in more traditional ways. In schools where more than one class was studying the same material (in some of the secondary schools), we were able to compare what students learned from world-building with what others learned from more traditional teaching. However, the VRRV project was by no means a controlled experiment and our findings to date are more suggestive than conclusive. We were also curious about how student characteristics affected how well they learned from world-building. We measured all students' general ability and middle and high school students' spatial ability. We also looked at how gender affected the outcomes. Pre- and post-tests of content were given to the secondary students. We also gave exit questionnaires to all students, interviewed some and videotaped others. The tests of content were teacher-built. They varied considerably in the content they tested, since at the secondary level each class built a different world in a different area of the curriculum. They also varied in scope and format. For this reason, we standardized test scores before conducting our statistical analysis. Finally, we gave a questionnaire that assessed students' attitudes towards science and computers. For the hors d''euvre, we gave every student an exit questionnaire. This asked them to rate on a five-point scale everything from how much they enjoyed the experience, to how easy it was to navigate in the virtual world, to the extent to which they experienced "presence" -- the conviction that they were really in another place when in the virtual world. We have gathered a large amount of data. The following is a brief description of some of the most important findings -- the tip of a very big iceberg. Beyond the purely descriptive data, the results that we mention below come from statistical analyses whose findings were significant at the .05 probability level. The Hors d''euvre Group Results From questionnaires given to 1,001 students in elementary schools, 922 students in middle schools and 949 high school students who experienced VR as part of the hors d''euvre: Students rated their enjoyment of the experience as very high with a negligible number of reports of queasiness. The latter finding is important because it allays the fear that exposure to VR, at least over short periods, has bad side effects. Difficulty moving around the world and interacting with it, as well as ratings of disorientation inside and on leaving the virtual world, decreased with age. The wand was difficult to hold and manipulate if you had small hands! Rated enjoyment of the experience and the sense of presence decreased with age. Perhaps the older children were more jaded in their attitudes to the technology. A factor analysis of the scale scores on the questionnaires showed remarkably consistent results across elementary, middle and high school students. In each case, the VR experience was shown to have three clear dimensions: Enjoyment/presence, disorientation/ malaise, and the ability to perform tasks. Interestingly, these three factors -- identifying affective, physiological and cognitive dimensions to the VR experience -- have been found, with very different populations, in other HITL projects. The Entrée Group Results From data obtained from 365 middle and high-school students who took part in world-building, not all of whom completed all tests, we can report: Students who built virtual worlds did learn the content they were expected to. Students who built virtual worlds did equally well regardless of their general ability. In the case of students who learned equivalent content in traditional ways, low-ability students did less well than high-ability students. This suggests that knowledge construction through a world-building activity helps students who might not learn well from traditional methods. This effect was most noticeable for boys. Students who built virtual worlds had consistently better attitudes towards science and computers after the experience, stating more frequently, for example, that scientists were honest, that they would consider taking science and computer courses in college, and follow careers in science and computing. Students learned more and enjoyed the project more who: used 3D models to visualize their world before they built their object on the computer; were easily able to find the object they had made when they visited their virtual world; and reported experiencing a high level of presence.
Students who had difficulty navigating in the virtual world or who lacked a clear understanding of the tasks they were to perform in it learned less and enjoyed the experience less. Students who collaborated with other students learned more and enjoyed the project more. Girls with low spatial ability reported feeling less presence. At the elementary level, boys reported they had learned more about VR than girls and also that they needed less time than girls to build their worlds. At the secondary level, boys enjoyed the world-building more than girls and spent more time on the computers. We can report, anecdotally, that boys often tended to "hog" the computers and to dominate a lot of the activities, which could account for these findings. High spatial ability was correlated with enjoyment, learning and the feeling of presence. Suggested Conclusions Our findings suggest a number of conclusions, all of which require confirmation under more rigorously controlled conditions. First, it worked! Our students did learn from designing and building virtual worlds. Notice that we do not claim that visiting the worlds they created taught them content. (We have learned in other projects, however, that students can learn from worlds built for them by someone else.) The acts of researching, designing and constructing worlds was the deciding factor. Next, the project seemed to level the playing field for less-able students. We believe that VR takes advantage of visual and kinesthetic abilities and learning strategies that are less likely to be used by students in more traditional, symbolic modes of instruction. Next, working with VR is motivating, as one would expect. Finally, general ability, spatial ability and gender all influence the cognitive and affective outcomes of working with VR. What We Plan to Do Next From now until the end of the academic year, the hors d''euvre part of the project moves to Nebraska. With further funding, we hope to take the project to other states in subsequent years. Of course, we are looking forward to the day when the entire project can be conducted over a network and we can dispense with the vans. But, again, to take that step requires significant resources that we do not yet have. World-building will continue in some Washington schools where the project's research on the effectiveness of world-building and of world-visiting will continue. Our studies will be more carefully designed and executed so that we can begin to explore more thoroughly the questions concerning what aspects of VR contribute to what kinds of learning, and the influence of general ability, spatial ability and gender on learning from building and visiting worlds. Further Information: Several reports about our work on educational applications of VR can be found at HITL's Web site (www.hitl.washington.edu) under "Publications." Information about HITL's Learning Center and the VRRV project can be found under "Projects"; information about our team is under "People." We will continue to make our reports available on the Web as we complete them. In the meantime, we are encouraged by the progress we report in this article and are looking forward to "pushing the envelope" with the help of the creative and energetic children in our the public schools here and around the country. William Winn, a professor in the Educational Technology program in the College of Education, is also Director of the Human Interface Technology Laboratory's Learning Center at the University of Washington. He holds an adjunct professorship in Technical Communication in the College of Engineering. Winn's research focuses on how people perceive and learn from maps, diagrams and pictures and on how cognitive and constructivist theories of learning can be used effectively by instructional designers. His work on the applications of Virtual Reality to education, some of which is reported here, brings together these two research interests. The VRRV project started while Winn was on sabbatical at the Human Interface Technology Laboratory during part of the 1994-95 school year. E-mail: firstname.lastname@example.org Products mentioned in this article: Provision 100 system, dVISE software; Division, Inc., Chapel Hill, NC, (919) 968-7797 Macromodel 3D; Macromedia, San Francisco, CA, (415) 252-2000
This article originally appeared in the 12/01/1995 issue of THE Journal.