Embodied AI Robots Enter Factories and Warehouses: What Else Must Robots Compete On After Smarter Brains?

At the CeMAT exhibition in Shanghai on November 3, robotics companies showcased embodied AI robots for logistics tasks including shelf climbing, four-directional bin retrieval, narrow-avoidance navigation, and multi-agent collaboration. Companies like Cainiao, Figure, Xindong Jiyuan, Atomix, and Xiangong Intelligent demonstrated hardware upgrades such as curved-path turning to save about two seconds per turn, dual-depth storage increasing density by 20%, and hybrid human-robot workflows where robots handle standard items while humans manage exceptions. The report notes that while software models have advanced, hardware constraints like gripper cost (up to one-third of total robot price), thermal limits, and chip power requirements remain barriers to widespread commercial adoption, with some specialized equipment offering faster ROI than humanoid alternatives at current prices.