With the rapid advancement of artificial intelligence, technologies like autonomous driving and embodied intelligence are revolutionizing industries. Visual perception, a key aspect of these intelligent systems, remains challenging in dynamic and unpredictable environments.
In real-world scenarios, intelligent systems must process vast data and handle extreme events like sudden dangers, drastic light changes, and strong flash interference. Traditional visual sensing chips struggle with distortion, failure, or high latency due to constraints like the “power wall” and “bandwidth wall,” impacting system stability and safety.
To tackle these challenges, Tsinghua University’s Center for Brain Inspired Computing Research (CBICR) developed an innovative vision sensing technology. They proposed a complementary sensing paradigm with a primitive-based representation and two visual pathways, inspired by the human visual system. This approach, detailed in the May 30, 2024, issue of *Nature*, decomposes visual information into primitive representations, mimicking human visual features to create two complementary visual pathways.
CBICR’s breakthrough led to the creation of the Tianmouc chip, the world’s first brain-inspired complementary vision chip. Tianmouc captures visual information at 10,000 frames per second, with 10-bit precision and a high dynamic range of 130 dB, while reducing bandwidth by 90% and maintaining low power consumption. This chip overcomes traditional visual sensing limitations, efficiently handling extreme scenarios to ensure system stability and safety.
The Tianmouc chip, combined with high-performance software and algorithms, was tested on a vehicle-mounted perception platform. In various extreme scenarios, the system demonstrated low-latency, high-performance real-time perception, showcasing its potential for intelligent unmanned systems.
The development of Tianmouc marks a significant breakthrough in visual sensing technology, supporting advancements in autonomous driving and embodied intelligence. Along with CBICR’s existing brain-inspired computing technologies like Tianjic, the Tianmouc chip enhances the brain-inspired intelligence ecosystem, driving progress toward artificial general intelligence.
