Chinese battery manufacturer CATL has begun large-scale deployment of humanoid robots in its battery pack production lines. On December 17, the company announced that its Zhongzhou facility is now operating the world’s first embodied AI humanoid robots in new energy battery pack manufacturing. Named “Moz,” each robot handles critical end-of-line (EOL) and direct current resistance (DCR) testing processes, which involve precisely connecting high-voltage test plugs to battery packs before shipment. These tasks traditionally carried spark risks and introduced variability in efficiency and quality when performed manually.
Moz is powered by end-to-end vision-language-action (VLA) models, giving it advanced environmental perception and task generalization capabilities. CATL reports that the robot excels in three core areas: precision adaptation, flexible operation, and efficiency. It independently compensates for deviations in material position and connection point changes, continuously adjusting its posture in real time. When handling flexible wire harnesses, Moz dynamically controls insertion and removal forces to ensure secure connections without damaging components. In production trials, it has sustained a connection success rate above 99%, matching the throughput of skilled human operators.
“Moz has become an indispensable member of our production line,” CATL stated. “Beyond its primary tasks, it autonomously detects wire harness connection status, reports anomalies immediately to reduce defect rates, and proactively switches to inspection mode between operations.”
Developed by Spirit AI, a robotics company within CATL’s ecosystem, Moz operates on CATL’s self-developed battery modules, showcasing successful collaboration along the supply chain. To translate production requirements into engineering specifications, CATL coordinated multiple departments for in-depth research on its production lines. This cross-functional effort led to a development plan that balances innovative vision with practical implementation, paving the way for broader adoption of embodied AI in intelligent manufacturing.
Source: Car News China
