Taiwan university develops light-controlled chip that mimics brain functions

台灣的大學研發出可模擬人腦功能的光電晶片

TAIPEI (Taiwan News) — A team led by National Yang Ming Chiao Tung University Photonics Department Professor Liu Po-tsun (劉柏村) has developed a light-controlled chip capable of simulating brain functions.

The research was published in the international journal Small. The innovation holds potential for applications in smart medical diagnostics, autonomous vehicle vision systems, wearable sensing devices, and robotics, per CNA.

Liu said the chip features a metal-based structure that mimics the brain's learning and memory processes. It demonstrates strong responsiveness to visible light. By applying light pulses and voltage control, the team successfully enabled the chip to emulate the conversion of short-term memory into long-term memory.

Beyond memory, Liu noted the chip also has sensing capabilities. This feature could help autonomous vehicles capture critical visual data or assist imaging devices like X-ray and CT scanners in identifying abnormal areas.

The team also created a small chip module that uses light signals to mimic how the brain learns and remembers. This module functions similarly to the human eye's retina by sensing varying brightness and colors in images, and it improves through learning; notably, it can retain learned information even after the light source is removed.

To test the chip's effectiveness, the team integrated it into an artificial neural network simulation platform to evaluate digit and image recognition. Even under challenging conditions, such as blurry images or interference from extraneous lines, the system achieved high recognition accuracy.

In 2023, the university’s Biomedical Engineering Department developed a graphene-based chip capable of monitoring and stimulating nerves. The department also collaborated with Abbott to implant the chip into the brains of lab animals for deep brain stimulation. Results showed the technology significantly improved depressive behaviors in the animals.