Chinese researchers have developed a neural chip that helps robots monitor and respond to movement instantly, with a speed four times faster than the human eye. In detail, a research team from Beihang University and Beijing Institute of Technology successfully designed a chip inspired by the lateral geniculate nucleus, a brain region located between the retina and the visual cortex that acts as a relay and filter, allowing the human visual system to focus its processing energy on fast-moving objects. Traditional robotic vision systems rely on cameras that capture still frames and track motion by changes in brightness between frames, but this method takes over half a second to process one frame, a delay that could cause disasters in applications like self-driving vehicles at high speeds. According to 'Indian Express', the researchers designed a specialized neural unit that monitors changes in light over time, enabling robotic vision to process motion instantly and focus its processing power on areas where motion occurs. Experiments conducted in driving simulations and using robotic arms to perform complex tasks showed that the chip reduced processing delay by about 75% and doubled motion tracking accuracy compared to previous methods. The research team confirms that the study raises video analysis speed to levels exceeding human capability by applying brain visual processing principles to a semiconductor chip, and it can be employed in preventing collisions in self-driving cars and instant object tracking in drones, as well as in fields requiring robots to read and respond instantly to human gestures. The chip faces some challenges in visually cluttered environments where multiple simultaneous movements occur, as it relies on optical flow algorithms to interpret the final image, but it could be useful in home environments where robots need to monitor small changes like gestures and facial expressions, making human-robot interaction more natural.
