While a standard hardware set allows apparatus to be shared amoung the members of the community, each user is encouraged to customize his wearable computer. The result is a constantly changing and improving environment. This section gives a brief overview of the systems used for the work in this paper, but for detailed or recent information, the reader is encouraged to visit the MIT wearable computing web site [Starner et al., 1995].
A standard system consists of a Private Eye display, Twiddler one-handed keyboard, PC-104 based 50MHz 486 computer, 16M of RAM, and approximately 1G of hard disk. The PC-104 3.6'' by 3.8'' board standard allows each user to add off-the-shelf 16-bit sound boards, video digitizers with on-board DSP's, PCMCIA adaptors, higher end 586 processors, or extra communications ports as desired. Cameras, biosensors, CRT or LCD displays, extra disk capacity, and custom clothing are also incorporated. Network connectivity in the region is achieved either through commercial digital cellular data service or a custom wireless system supported by the second author. When on-board processing power is not sufficient for analyzing video for the above techniques, a self contained, full duplex amateur television transmitter/receiver (Figure 15) is used in conjunction with HP or SGI workstations. Video is transmitted from the head mounted camera, analyzed remotely, and returned to the user's display with appropriate graphics overlaid on the image [Mann, 1994]. This method emulates the experience of having significant processing power locally and allows rapid prototyping. As techniques are proven, the code is optimized and moved to the processor(s) of the local wearable computer, if possible. Thus, concept demonstrations are gradually integrated into everyday use. As of this writing, the face recognition code has been ported, and ports of the finger tracking and visual tag code are underway.
Figure:
Remote video processing system.