Hugo de Garis of ATR (Kyoto) believes that evolvable hardware (EHW) based on FPGA chips will become a billion dollar industry. At present, he counts only three researchers working on it. Tetsuya Higuchi, at ETL, has $1M from MITI to develop an evolvable chip by 3/97. (Automatic welding is one of his test domains.) Hitoshi Hemmi, at ATR, uses genetic programming to develop functions in software, then downloads them to programmable gate arrays. Adrian Thompson, at Sussex University, is the only person modifying the FGPA configuration with each evolutionary step. One advantage of EHW is that you don't need to synchronize to an external clock pulse: the logic circuits will form around the time constants of the hardware substrate. (Inman Harvey argues that the digital domain is too spikey, and that analog functions should be used instead.) One test application by Thomson was an oscillator with a period in milliseconds, evolved from nanosecond components. FGPA reprogramming currently takes 30 minutes, but Xilinx is about to introduce an $80 Series 6000 FPGA that will be much faster. Thompson intends to evolve control links that will be faster than the software provided by Xilinx. The next EHard workshop/conference will be in Japan in 10/96. [, Evolutionary Hardware Workshop Report, 11/7/95. John Koza , genetic-programmers, 12/1/95. Bill Park.]

There is now a web page for neuro-fuzzy VLSI implementations and their application to real-time control, at . [Marcello Chiaberge, comp.ai.fuzzy, 11/2/95.]

If you have questions on genetic programming, see the GP home page, FAQ, papers, bibliography, code archive, and links at . Files can be FTP'd from . [Adil Qureshi , genetic-programming, 8/22/95. Bill Park.]