Autonomous Robot Uses Pure Vision Edge Brain System for Navigation

People's Daily reports that a white robot featuring large eyes and a Mars-rover-styled chassis demonstrated autonomous navigation at an industrial exhibition in Shanghai. Without relying on preset maps or remote controls, the machine successfully avoided pedestrians and located paths solely using its vision system. Named Danbao, the robot carries a pure-vision edge brain developed jointly by Fudan University's School of Future Information Innovation and the Jimu Tianshu team. Research team leader Zhang Tianzi explained that robots traditionally depend on remote operations, preprogramming, or pre-mapping rather than real-time understanding of their surroundings. The system treats all physical movements as navigation tasks, framing locomotion as moving the body to a destination and manipulation as guiding fingertips toward target objects. While many units run separate systems for walking and grasping, this unified edge brain coordinates full-body motion and powers physical navigation skills. Zhang stated that visual spatial intelligent navigation features low computing power requirements and high intelligence. The model operates on edge hardware within 20 watts of power while managing up to 8 visual inputs, allowing the robot to make autonomous decisions without preset maps. The development team is actively adapting this technological framework to several machine types, including delivery, inspection, agricultural, and cleaning robots. Future work aims to expand capabilities from autonomous movement to full task execution and intention understanding.