The original VuMan LTI task was implemented using a custom hypertext system, primarily because of processing constraints imposed on that device (a 25MHz Intel 386). As we were primarily interested in the speech interaction aspects of the task, we chose to implement our system using a standard notebook computer with a more powerful processor. The computer, plus a battery power supply and control hardware for the head-mount display were placed in a pack worn on the user's back. This arrangement, although bulkier than the VuMan package (which can be attached to the user's belt), allowed users to freely move about, inspect the underneath of the vehicle, climb to the roof, etc.
For purposes of the current study the task was recast as a hypertext document using standard html format, allowing for a more rapid and flexible design process. The html/http framework offers a simple yet powerful mechanism for unifying information resources useful for this task, both for data collection and for access to distributed resources. Using a standard browser also allowed us to incorporate a variety of information, such as a scanned repair manual and video clips keyed to individual steps in repair procedures, all accessible by voice.
Figure 1: The SPEECHWEAR system