Aspinity Inc. (Pittsburgh, Penn.), a supplier of low-power analog AI processors, has closed a US$5 million round of funding that included Korean distributor of automotive semiconductors, Unitrontech.
The Series B round brought the amount raised by Aspinity since its founding in 2015 to US$19 million. It also included previous investors Anzu Partners, Birchmere Ventures, Mountain State Capital and Riverfront Ventures.
Aspinity has specialized in “always-on” applications and the analysis of real-time time-series data, such as sensor data. Aspinity’s first product, the AML100 analog machine learning core keeps the MCU and other digital processors asleep until meaningful data are detected. The software-programmable chip can address many different inferencing applications with Aspinity’s Python-based SDK that allows development, testing, and verification of analog machine learning models.
Aspinity and Unitrontech intend do apply the technology to near-zero power after-market and integrated AI solutions for automotive applications. For example, an always-on dashcam solution that leverages sensor fusion and AI to detect and record only the events that are important to the car owner. In this application, the AML100 can be programmed to detect impacts and intrusions in general, or to detect and issue alerts on specific events such as glass break.
“Our strategic partnership with Aspinity reinforces our commitment to providing the most innovative solutions to our automotive customers,” said Richard NamKoong, CEO of Unitrontech, in a statement.
Tom Doyle, CEO and co-founder of Aspinity, said: “No other AI chip can interface directly with analog sensors and inference while the data are still in their native analog domain, keeping downstream computing resources focused solely on relevant data. This is the key to minimizing power consumption and to a future of ubiquitous AI systems.”
Related links and articles: