RoadNarrows
 Robotics

This is THE CogniMem chip, a neural network chip for pattern learning and recognition w ith 1024 neurons working in parallel. It is both a powerful model generator and non-linear classifier and the unique parallel architecture of its neurons offers two technological breakthroughs: a recognition time independent of the number of neurons in use and the ability to cascade multiple CogniMem chips to size the network at will. Chip control can be made through a parallel or I2C port.

This is a breakout board containing the CM-1K chip on with standard headers for developing prototypes with the CM-1K chip directly.

he CM-IR1K Image Recognition Board offers developers and OEMs a comprehensive platform to evaluate the CogniMem neural network for image recognition applications. The board features a Micron CMOS sensor, a CogniMem chip with 1024 neurons, an Actel FPGA accessible to programmers and a Flash memory to save the application settings and knowledge.

The CM-EB1K Evaluation Board offers developers and OEMs a comprehensive platform to evaluate the CogniMem neural network. The board features a CogniMem chip with 1024 neurons, an Actel FPGA accessible to programmers and a digital input bus for easy connectivity to external sensors.

CM-E2K Expansion Board includes 2 CM-1K giving 2048 neurons and can directly connect to a digital video or signal source (8bits data width). CM-E2K can be configured to be accessed by the CM1K parallel bus or by the I2C two wires bus. Up to four CM-E2K can stacked on the top of each other, totaling 8192 neurons.

The Vision Sensor Board offers OEMs a comprehensive platform to develop systems for machine vision. The board features an Aptina CMOS sensor, CogniMem chips with 1024 neurons, an Actel IGLOO FPGA accessible to programmers and a Flash memory to save the application settings and knowledge. User I/Os are implemented on the board in a flexible variety of configurations, including an I²C serial bus, an RS232 bus and 8 general purpose I/Os brought to header pins.