There is a small (large) problem with video recognition, even with the ER1 software. If you pose for a picture, the ER1 will recognize you and store the picture, but if you appear in front of the camera wearing a different shirt you chances of being recognized grows very small. Now some people might say well that is ER1's problem, but let me give you a little background on the company. The makers of ER1 have at this time the most advance video software available anywhere.
Their software is so good that Sony bought a license to use it in their Aibo robotic dogs.
Anyone here can go back a page or two and find all kinds of posts about the ER1 made a few months ago. Some of the information is very useful, however, interfacing the video software with Hal then becomes a master programmers problem. Such a program could spend most of its time just looking around and trying to find or identify everything that it sees. That could lead to very litter interaction with the rest of the program and might even defeat the main purpose of Hal, which is to interact with its user.
Of course, if you could use two computers which could communicate with each other, then you might just simplify the problem.
Bill