hlr@well.sf.ca.us (Howard Rheingold) (04/07/90)
philbo@chaos.cs.brandeis.edu (Phil Gross) writes: >Without more than visual and aural feedback, VR is *never* going to be >as good as the real thing. U-F seems to be just a miniature motion >detector, or a 3-d version of those old touch screens. Do you >remember those? They had a slightly recessed screen, and lots of >holes where beams of indeterminate nature were sent parallel to the >screen a few milimeters away. A 3d infared box that can do this is no >big deal, the important value is that they can afford to market it so >cheaply. >I disagree that this infared technology has much use in a virtual >reality/cyberspace setting. One of the biggest advances in the last >few years in vr research is the use of the VPL data-glove, a glove that >not only senses motion, but can simulate textures as well... I visited VR research facilities at ATR, MITI, Fujitsu, and NTT recently. People at ATR and NTT are aiming at "wireless" VR. They showed me some demos of a system that involves video, machine vision and neural network software. A video camera tracks the user. Pattern detection software attempts to track the user's gestures. Neural networks help the system learn the user's gestural repertoire. They have a way to go. Does anybody who is knowledgeable about machine vision and neural networks have an opinion about how well such a system might work? I've played with the dataglove a fair amount, and unless they have come up with something startling and new, the glove does NOT simulate textures. Margaret Minksy at Media Lab and UNC has been using a force-reflective joystick with her own software that does a remarkable job of simulating texture. But it is a joystick, not a glove.