Learning edge AI chip slashes power for the IoT
Rohm has developed a low power edge AI processor that includes learning as well as inference for predictive maintenance in the Internet of Things (IoT)
The prototype BD15035 AI chip is based on a three-layer neural network AI circuit developed by Professor Hiroki Matsutani of Keio University. Rohm reduced the AI circuit from 5 million gates to just 20,000 gates to act as a proprietary AI accelerator.
This AxlCORE-ODL block has been combined with Rohm’s high-efficiency 8-bit CPU tinyMicon MatisseCORE IP to enables both AI learning and inference for predictive maintenance of motors and sensors in real-time.
The combination of the 20,000 gate AI core and the 8bit controller cuts the typical power consumption to around 30mW, a tenth that of other edge AI processors.
- STM32 wireless module enables predictive maintenance
- Edge AI training on microcontrollers boosts IoT predictive maintenance
- NB-IoT module supports predictive maintenance
This allows real-time failure prediction in a wide range of motor and sensor applications to provide an anomaly detection score for unknown input data for equipment without involving a cloud server. This allows equipment to be examined and for maintenance to be scheduled before a failure occurs, preventing costly downtime.
Rohm says it plans to incorporate the AxlCORE-ODL AI accelerator into various other chips for motors and sensors. Commercialization is scheduled to start in 2023, with mass production planned in 2024.
“As IoT technologies such as 5G communication and digital twins advance, cloud computing will be required to evolve, but processing all the data on cloud servers is not always the best solution in terms of load, cost, and power consumption,” said Professor Matsutani of the Dept. of Information and Computer Science, Keio University, Japan.
“With the ‘on-device learning’ we research and the ‘on-device learning algorithms’ we developed, we aim to achieve more efficient data processing on the edge side to build a better IoT ecosystem. Through this collaboration, ROHM has shown us the path to commercialization in a cost-effective manner by further advancing on-device learning circuit technology. I expect the prototype AI chip to be incorporated into Rohm’s IC products in the near future.”
The 8bit tinyMicon MatisseCORE controller has an instruction set optimized for embedded applications together with the latest compiler technology to deliver fast arithmetic processing in a smaller chip area and program code size. High-reliability applications are also supported, such as those requiring qualification under the ISO 26262 and ASIL-D vehicle functional safety standards, while an proprietary onboard ‘real-time debugging function’ prevents the debugging process from interfering with program operation, allowing debugging to be performed while the application is running.
Rohm has developed an evaluation board with Arduino-compatible terminals that can be fitted with an expansion sensor board for connecting to a microcontroller. Wireless communication modules for WiFi and Bluetooth are mounted on the board along with 64kbit EEPROM memory. Connecting sensors and attaching them to the target equipment allows the board to verify the effects of the AI chip from a display. This evaluation board will be loaned out from Rohm’s sales team in Europe.
Other IoT predictive maintenance articles
- Xerox completes spin out of Novity predictive maintenance tech
- Predictive maintenance at the edge cuts bandwidth
- ABB develops hybrid predictive maintenance technology
- Sensirion buys AI predictive maintenance firm
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
