A team from Tsinghua University's brain-like Computing Research Center recently developed the world's first brain-like complementary vision chip, "Tianmou Core," which was published as a cover article in the international academic journal Nature on May 30. In the open world, intelligent systems not only have to deal with huge amounts of data, but also need to deal with extreme events such as sudden dangers in driving scenes, drastic light changes at the mouth of the tunnel and strong flash interference at night, said Shi Luping, corresponding author of the paper and professor of the Department of precision Instruments at Tsinghua University. In the face of such scenes, traditional visual perception chips often have distortion, failure or high delay, which limits the stability and security of the system. In order to better address the above problems, a team from the Research Center of brain-like Computing at Tsinghua University focused on brain-like visual perception chip technology and proposed a new paradigm of brain-like visual perception based on complementary dual-pathway visual primitives. The picture shows "Tianmou Core". (Department of Precision Instruments, Tsinghua University) "This paradigm draws on the basic principles of the human visual system, disassembles the open world visual information into information representations based on visual primitives, and through organic combination of these primitives, imitates the features of the human visual system, forming two complementary advantages and complete information visual perception pathways." Shi Luping said. Based on this new paradigm, the team further developed the world's first brain-like complementary vision chip "Tianmaicin", which realizes high-speed, high-precision and high-dynamic range visual information acquisition at the cost of extremely low bandwidth and power consumption, and can effectively cope with various extreme scenarios to ensure the stability and security of the system. At the same time, based on the "Tianmou core", the team also independently developed high-performance software and algorithms, and conducted performance verification on the open environment vehicle platform. In a variety of extreme scenarios, the system realizes real-time perceptual reasoning with low latency and high performance, showing its application potential in the field of intelligent unmanned systems. Zhao Rong, the corresponding author of the paper and a professor at the Department of precision Instrumentation at Tsinghua University, said that Tianmou Core opens up a new way for important applications such as autonomous driving and embodied intelligence. Combined with the team's technology accumulation in the application of brain-like computing chip "day movement", brain-like software tool chain and brain-like robot, the addition of "Tianmouxin" will be able to further improve the brain-like intelligence ecology and effectively promote the development of artificial general intelligence. According to reports, this is the second time that the team has appeared on the cover of Nature after the heterogeneous fusion brain-like computing "Day movement", marking a basic breakthrough in both directions of brain-like computing and brain-like perception.