Three teams of corporate bigs are vying to provide the technology for the next generation of television. The Federal Communications Commission was expected to pick a winner early this summer, paving the way for one team to rake in huge royalty payments. Now, it looks as if a decision won't come until early 1994. That's because the three teams--one led by Thomson, Philips, and NBC, another by General Instrument, and a third by Zenith Electronics--are closeted in negotiations to form what's being called a "grand alliance" merging the best aspects of each of their high-definition television (HDTV) systems. An agreement might be reached by early April.
The teams have some major incentives to get together. Not only would each cut the risk of an outright loss, but an agreement would forestall legal protests from losing teams that could delay things even more. If the deal comes off, it would take most of 1993 to build and test the hybrid system. That means U.S. consumers probably wouldn't get their first taste of HDTV until late 1994 or early 1995.
Researchers at the University of California at Berkeley are taking a new approach to the robotic eye with a multipurpose-chip design they call the Cellular Neural Network Universal Machine. It will have hundreds or thousands of tiny, neuronlike processors that work on real-world data in their native, analog form instead of changing them into ones and zeroes. That makes the chip faster than digital neural-network chips, such as Intel Corp.'s. And unlike other analog neural chips, such as one from Synaptics Inc. that reads numbers for check authorization, each processor has lots of its own memory. That means each processor on the chip can be connected to fewer neighbors, simplifying wiring.
So far, only small versions of the chips exist. But Berkeley computer scientist Leon O. Chua, who invented the Universal Machine last year with Tamas Roska of the Hungarian Academy of Sciences, is designing a bigger one that could be quickly reprogrammed for special tasks. Berkeley neurobiologist Frank Werblin speculates that a version in a decade or so could even serve as a primitive bionic human eye.