This paper is an overview of an enormous undertaking, both in terms of scale and implication. The “Brain Builder Group” of ATR Labs in Kyoto, Japan, headed by Hugo de Garis, is currently working on two highly related, but somewhat independent projects: the “CAM-Brain Project” and the “Robokoneko Project”. The CAM-Brain Project is primarily concerned with developing “evolvable hardware” capable of “growing” and maintaining a functioning artificial brain sufficiently powerful for real-time applications. On the other hand, the Robokoneko Project (“robokoneko” is Japanese for “robot kitten”) aims at developing a physical kitten-like robot for the artificial brain to control. Since both are being pursued in parallel by the same group, for convenience, I will henceforth refer to them as though they were a single project unless an explicit distinction is being made.
Although the project is primarily centered at ATR, much of the hardware development has been contracted out to a company named “Genobyte Inc.” (in Boulder, Colorado) under the supervision of Michael Korkin. In addition, there are around 100 other researchers worldwide involved with various tasks relating to Robokoneko and the CAM-Brain.
The Brain Builder Group has been underway for approximately 6 years and its originally stated goal was to have a functioning artificial brain of 1 billion artificial neurons by the year 2001. Although the artificial brain currently being developed will only have 40 million neurons, the Group is not far behind in its projected time line. They still hope that their robotic kitten will be completed by 2001.
Projects of this scope obviously address issues in a wide range of areas; section 2 provides a
summary of the various aspects of the project and their challenges, successes, and current states.
Section 3 discusses some of the implicit assumptions / philosophical approaches taken by the
researchers and the merits and problems of the project. Finally, section 4 concludes.
Agah, Arvin, Hugo de Garis, Michael Korkin, Katsunori Shimohara, and Sung-Bae Cho. ms. “Architectural and Functional Specifications for a Robot Kitten ‘Robokoneko’ to be Controlled by a 10,000 Evolved Neural Net Module Artificial Brain.”
Brooks, R. A. 1997. Intelligence without Representation. In J. Haugeland (Ed.) Mind Design II. pp. 395-420. MIT Press: Cambridge, MA.
de Garis, Hugo, Michael Korkin, Katsunori Shimohara. ms(a). “ATR’s ‘CAM-Brain Machine’ (CBM) and Artificial Brains: An FPGA Based Hardware Tool which Evolves a Neural Ner Circuit Module in a Second and Updates a 40 Million Neuron Artificial Brain in Real Time.” submitted to: Journal of VLSI Signal Processing Systems (JVSPS), Special Issue on Custom Computing Technology.
de Garis, Hugo, Andrzej Buller, Michael Korkin, Felix Gers, Eiji Nawa, Michael Hough. ms(b). “ATR’s Artificial Brain (‘CAM-Brain’) Project: A Sample of What Individual ‘CoDi-1Bit’ Model Evolved Neural Net Modules Can Do with Digital and Analog I/O.”
de Garis, Hugo, Nikolai Petroff, Michael Korkin, Gary Fehr, and Eiji Nawa. ms(c). “A 40 Million Neuron Artificial Brain for an Adaptive Robot Kitten ‘Robokoneko’”. submitted to: Connection Science Journal, Special Issue on Adaptive Robots.
de Garis, Hugo. 1993. “Evolvable Hardware: The Genetic Programming of Darwin Machines.” International Conference on Artificial Neural Networks and Genetic Algorithms. Innsbruck, Austria.
de Garis, Hugo. 1994. “CAM-Brain: Growing an Artificial Brain with a Million Neural Net Modules Inside a Trillion Cell Cellular Automata Machine.” Journal of the Society of Instrument and Control Engineers of Japan (SICE). Vol. 33, No. 2. Society of Instrument and Control Engineers of Japan.
de Garis, Hugo. 1997. “Evolvable Hardware: Principles and Practice.” Communications of the Association for Computer Machinery (CACM Journal).
Korkin, Michael, Norberto Eiji Nawa, and Hugo de Garis. 1998. “A ‘Spike Interval Information Coding’ Representation for ATR’s CAM-Brain Machine (CBM).” Int. Conf. on Evolvable Systems. Lausanne, Switzerland.
Gers, Felix, Hugo de Garis, and Michael Korkin. 1997. “CoDi-1Bit: A Simplified Cellular Automata Based Neuron Model.” AE97 Artificial Evolution Conference. Nimes, France.
Levy, Steven. 1992. Artificial Life. Vintage Books: New York.
Thelen, Esther. 1995. “Time-scale dynamics and the development of an embodied cognition.” In Port and van Gelder (eds). Mind as Motion: Explorations in the Dynamics of Cognition. Bradford Books, MIT Press: Cambridge, MA. pp. 69-100.
Hugo de Garis and ATR (including all CAM-Brain and Robokoneko project papers listed above): http://www.hip.atr.co.jp/~degaris/
Genobyte Inc.: http://www.genobyte.com/
Xilinx Inc.: http://www.xilinx.com/
Working Model 3d: http://www.workingmodel.com/