ROHM has created an on-device learning AI chip (SoC with on-device learning AI accelerator) for edge computer endpoints in the IoT field. It uses AI to predict failures in electronic devices fitted with motors and sensors in real-time offering ultra-low power consumption.
Generally, AI chips conduct learning and inferences to accomplish AI functions, as learning requires a considerable amount of data to be captured, compiled into a database, and updated as necessary. So, the AI chip that performs learning needs substantial computing power that consumes a large amount of energy. Until now, it has been challenging to develop AI chips that may learn in the field consuming low power for edge computers and endpoints to create an efficient IoT ecosystem.
Based on an 'on-device learning algorithm' created by Professor Matsutani of Keio University, the company's newly developed AI chip mostly comprises an AI accelerator (AI-dedicated hardware circuit) and its high-efficiency 8-bit CPU 'tinyMicon MatisseCORE'. Merging the 20,000-gate ultra-compact AI accelerator with a high-performance CPU facilitates learning and inference with ultra-low power consumption of only a few tens of mW (1000× smaller than traditional AI chips capable of learning). This permits real-time failure prediction in a broad variety of applications since 'anomaly detection results (anomaly score)' can be output numerically for unknown input data at the site where equipment is installed without involving a cloud server.
Into the future, the company plans to include the AI accelerator employed in this AI chip in assorted IC products for motors and sensors.
Professor Hiroki Matsutani, Dept. of Information and Computer Science, Keio University, Japan, said: "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. 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."