NeuroMem Technologies has launched the NeuroTile “mobile AI” module for wearables and robotics. NeuroTile is packaged with Arduino IDE code, and is pin-compatible with Intel Curie modules. It combines a low-power STM32 microcontroller, Lattice FPGA, an NM500 neuromorphic chip on a 35 mm x 26 mm board. An additional STM32 microcontroller delivers extended software capabilities and Bluetooth Low Energy (BLE) support.
NeuroTile leverages NeuroMem’s pattern recognition and training technology to deliver real-time AI inferencing at low power consumption. The NM500 neuromorphic chip features consist of:
- 576 neurons (up to 256 bytes each)
- 85 Gigaoperations per second @ 37 MHz and 120 MW
- Constant 8 µs learning and recognition time
- Expansion bus to additional neuromorphic chips
- Unsupervised learning capability
“For the first time, developers can use our disruptive AI technology combined with multiple sensors to create breakthrough applications easily and in a very short time,” says Pierre Brunswick, Chairman of NeuroMem Technologies. “This goes beyond software-based deep learning, and creates learning capabilities for AI hardware that will learn as its applications dictate.”
NeuroTile is part of the company’s NeoCurie line of “mobile AI” products, which also includes the TinyTile module.
For more, visit www.general-vision.com/technology.eletter-09-11-2017