Collaborative virtual environments (CVEs) have been proposed for a wide range of applications, including reviewing design in manufacturing industry, training astronauts, simulating the ergonomic aspects of handling materials in a factory, and visualizing and interpreting data in the oil and gas industry. Multiple users in CVEs can cooperate at any of three levels. On the first level, users coexist in the CVE and can perceive and communicate with each other. On the second level, each user can individually modify the contents of a scene. Only on the third level can users simultaneously act on the same object and jointly perform operations such as changing an object's position and orientationwhich we take for granted in the physical world. Level 3 cooperation therefore is highly desirable for many CVE applications.
To accommodate cooperative manipulation, a CVE's interface software must integrate the inputs of different users according to some predefined rules of interaction, resolve conflicts, and ensure that the resultant manipulations are reflected in the scenes that are rendered to each user's display. The situation is complicated by the fact that many CVEs will provide only visual information to users, and haptic feedback is unavailable because of technological constraints (for instance, the need for a large working volume) or cost.
This paper describes a behavioral experiment that investigated the effect of two substantially different rules of interaction when pairs of participants performed a ge-neral-purpose cooperative ma-nipulation task known as the piano movers' problem, which involves manipulating a bulky object through doorways and along corridors in CVEs (see Figure 1). One of the rules only allowed the synchronized component of participants'actions to take place, and the other manipulated the object according to the mean of participants' action and allowed them to interact asynchronously.
The most important findings were as follows: In terms of the total time participants took to complete the tasks neither rule was superior, but that did not tell the whole story. Detailed analysis of the data showed that parts of the task that required participants to perform similar actions were performed most quickly if the interface allowed only synchronized movement to take place. However, if participants needed to perform different types of movement (for example, one maneuvered the object through a door while the other traveled down the corridor), the asynchronous rule of interaction was superior. In the physical world people switch bet-ween synchronous and asynchronous interaction at a largely subconscious level. This research indicates the benefits that could accrue if CVE interfaces allowed both types of interaction and allowed users to switch between them as required. Simple versions of the piano movers' task were performed in a CVE almost as quickly as individual users did on their own, but the most difficult task that was studied took 50 percent longer in a CVE. This represents a substantial cooperation overhead (reduction in performance resulting from having to cooperate with another person, compared with doing something by oneself) and indicates the large improvements that could be made to CVE interfaces if force feedback can be implemented over a large working volume.
Roy A. Ruddle School of Computing, University of Leeds, UK
Justin C.D. Savage School of Psychology, Cardiff University, UK
Dylan M. Jones School of Psychology, Cardiff University, UK
Figure 1. Inside a CVE used in the study, as viewed by
one participant above and another participant below. The
participants interacted with the object via two virtual humans,
which were their embodiments in the CVE. The wireline
highlighting indicates that the object is colliding with one of
©2002 ACM 1072-5220/02/1100 $5.00
Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, to republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee.
The Digital Library is published by the Association for Computing Machinery. Copyright © 2002 ACM, Inc.