THIRUVANANTHAPURAM: The prompts that are fed artificial-intelligence platforms more often than not travel great distances to cloud-server farms and get processed there, before reaching you as responses. But what if the AI component of foundational systems, such as personalised health-monitoring frameworks and cameras on robots to detect external motion, is executed at the very place of origin?
In what could be a revolutionary solution to privacy concerns and help augment the country’s semiconductor capabilities, researchers from the College of Engineering Trivandrum (CET) and Netrasemi — a Thiruvananthapuram-based startup — have developed a chip, CNVR1K (CET-Netrasemi Vision SoC R1000), that can process AI requirements in compact devices.
Created as part of a project headed by college alumni, including CET faculty and Netrasemi staff, the CNVR1K is being touted as the country’s first AI-capable chip developed by an academic institution.
“The major concern often raised by people using AI-based technology is privacy. Why should the server, which is a third party, store my information, which could even be AI-enabled CCTV monitoring of my house? This is where our chip comes into play, by solving the AI element there itself, without depending on big server farms,” said Netrasemi CEO Jyothis Indirabhai, who is also one of the chief investigators of the project.
Designed locally and fabricated at the Taiwan Semiconductor Manufacturing Co, the chips will be sent to Bengaluru for packaging. It will be ready for commercial production within a year, researchers said. Carried out under the ministry of electronics and information technology’s Chips to Startup programme, the project received `4.7 crore in financial assistance over the three years of chip development.
“Designed with the capacity for handling AI and internet of things (IoT), R1000 can be incorporated in various projects. For instance, our chip can be used in a device to constantly monitor the ECG of a cardiac patient and inform technicians concerned if any anomaly is detected,” Jyothis said.
By processing data locally, the chip’s use of Edge AI reduces cloud energy consumption and cooling water requirements.
“Beyond classroom learning, our students received training on how chip designing takes place. Projects were divided among students, faculty members and junior research fellows of several departments. Today, recruiters are showing increased interest in hiring our students upon seeing their exposure to such initiatives,” Chief investigator Sreelatha G, who leads the CET team, said.
CET’s Adersh V R, Nikhil M, and Resmi E and Netrasemi’s Sreejith Varma C K and Deepa Geetha were also a part of the project as co-chief investigators.