Japanese electronics company Rohm has developed an on-device learning AI chip for edge computer endpoints in the IoT field.

It uses artificial intelligence (AI) to predict failures in electronic devices equipped with motors and sensors in real time with low power consumption.

Generally, AI chips perform learning and inferences to achieve artificial intelligence functions, as learning requires that a large amount of data gets captured, compiled into a database and updated as needed. So, the AI chip that performs learning requires a lot of computing power that necessarily consumes a large amount of power. Until now it has been difficult to develop AI chips that can learn in the field consuming low power for edge computers and endpoints to build an efficient IoT ecosystem.

Based on an on-device learning algorithm developed by Hiroki Matsutani of Keio University, the AI chip mainly consists of an AI accelerator and Rohm’s TinyMicon MatisseCore 8bit CPU.

“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 in terms of load, cost and power consumption,”: said Matsutani. “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 commercialisation 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.”

Combining the 20,000-gate compact AI accelerator with a high-performance CPU enables learning and inference with power consumption of a few tens of mW. This allows real-time failure prediction in a wide range of applications, since anomaly detection results can be output numerically for unknown input data at the site where equipment is installed without involving a cloud server.

Going forward, Rohm plans to incorporate the AI accelerator used in this AI chip into various IC products for motors and sensors. Commercialisation is scheduled to start in 2023, with mass production planned for 2024.

TinyMicon MatisseCore is Rohm’s proprietary 8bit CPU developed for making analogue ICs more intelligent for IoT ecosystems. An instruction set is optimised for embedded applications together with the latest compiler technology to deliver fast arithmetic processing in a smaller chip area and programme 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 the proprietary onboard real-time debugging function prevents the debugging process from interfering with programme operation, allowing debugging to be performed while the application is running.

The prototype AI chip is based on an on-device learning algorithm developed by Matsutani. Rohm downsized the AI circuit from five million gates to 20,000 to reconfigure for commercialisation as a proprietary AI accelerator controlled by the firm’s 8bit TinyMicon MatisseCore CPU that enables AI learning and inference with power consumption of a few tens of mW. This makes the numerical output of anomaly detection results possible for unknown input data patterns at the site where equipment is installed without involving a cloud server or requiring prior AI learning, allowing real-time failure prediction by onsite AI while keeping cloud server and communication costs down.

For evaluating the AI chip, an evaluation board is available equipped with Arduino-compatible terminals that can be fitted with an expansion sensor board for connecting to an Arduino MCU. Wifi and Bluetooth communication modules along with 64kbit EEPROM are mounted on the board, and by connecting units such as sensors and attaching them to the target equipment it will be possible to verify the effects of the AI chip from a display. This evaluation board will be loaned out from Rohm sales.

Rohm Semiconductor is a ¥452.1bn company with more than 23,000 employees. Production takes place in manufacturing plants in Japan, Germany, Korea, Malaysia, Thailand, Philippines and China.