Prof. Dr. Hugo de Garis
CAM-BRAIN MACHINE (CBM) FPGA DESIGN
The images that follow, show various aspects
of the FPGA based CBM circuits as designed by
Dr. Michael KORKIN of Genobyte Inc.
Click to enlarge
Layout of a 2x2x2 block of cells in a XC6216 FPGA
|
Layout of a 4x4x2 block of cells in a XC6216 FPGA
|
Fully placed and routed XC6216 with 64 CoDi cells
|
3D CA space with one |
Multimodular |
Layout of a CBM 3D cell |
CAM-Brain Machine (CBM)
The long term aim of STARLAB's CAM-Brain Project is to build artificial brains, e.g. a billion neuron artificial brain by the year 2001. Neural networks are based on cellular automata, and are evolved using a Genetic Algorithm (GA) at electronic speeds using the latest in FPGAs (field programmable gate arrays), i.e. Xilinx's XC6264 chips. CA based neural circuits can be grown and evalutated totally in hardware in microseconds, making possible a complete run of a GA (i.e. tens of thousands of circuit growths and evaluations (fitness measurements)) in less than a second.
Up to 64,000 evolved neural net modules can be assembled into humanly designed artificial brain architectures, and each CA cell in the whole brain of millions of cells (stored in RAM) can be updated (using the CBM) thousands of times a second, which is easily fast enough for real time control of robots. The specialist piece of hardware which performs the above functions is called a "CAM-Brain Machine (CBM)".
Construction
The CBM is now being designed and built by Dr. Michael KORKIN of Genobyte Inc., and should be finished by the end of 1999. The CBM will be able to update CA cells at the incredible speed of over 130 BILLION a second!
A kitten robot (ROBOKONEKO) will be designed in parallel in 1998 and beyond, plus a 64,000 module artificial brain. These 3 components will be assembled and tested in 2000+. The brain will control the kitten like behaviors of the kitten robot. With 64,000 modules, the kitten robot should be remarkably life like in its behaviors.
If the CBM is successful, it will be revolutionary for the following reasons.
A) It will be the world's first significant example of (functional-level, as opposed to gate-level) "E-Hard" (Evolvable Hardware), an idea de Garis had in 1992.
B) It will "blow away" the traditional approach to neural networks (i.e. essentially using a single net, with tens of neurons). The CBM will allow tens of thousands of modules to be assembled into artificial brains, i.e. tens of millions of artificial neurons. The CBM will render traditional neural net papers "unsexy", because real brains have many neurons as opposed to tiny artificial networks studied by most neural net researchers today.
If the CBM, Robokoneko (kitten robot) and the 64,000 module brain architecture are successful in combination in 2000+, then the CAM-Brain Project will be revolutionary for the following reasons.
A) It will create a new field, called "Brain Building", a major step forward in the field of AI (Artificial Intelligence).
B) If Brain Building is successful, then companies will get very interested, creating a trillion dollar world industry within 20 years, i.e. building "brain like" computers.
A paper on the CBM can be found in (.ps.gz format) or (.pdf format). (See "papers" page for .pdf instructions).