Shanghai (Gasgoo)- Recently, CATL officially commissioned what it describes as the world's first new energy battery PACK production line to deploy humanoid embodied-intelligence robots at scale, at its Zhongzhou base, according to a press release CATL issued via its WeChat account on December 18. The humanoid robot, known as "Xiao Mo," is now capable of accurately handling complex tasks such as connecting battery interfaces, a development widely seen as a milestone for the real-world application of embodied intelligence in smart manufacturing.
The processes replaced by Xiao Mo—EOL and DCR—represent the final functional testing stage before a battery pack leaves the production line. Because this stage involves multiple product variants, small batch sizes, and high flexibility requirements, it has traditionally relied heavily on manual labor. Workers must precisely connect test plugs carrying several hundred volts of high voltage to designated points on the battery pack, a task that carries safety risks and has long posed challenges in terms of efficiency and consistency.
The introduction of Xiao Mo has fundamentally reshaped this workflow. Powered by an advanced end-to-end vision–language–action (VLA) model, the robot demonstrates strong environmental awareness and an ability to generalize across tasks, allowing it to operate reliably in complex and variable production conditions.
In practice, Xiao Mo can independently respond to uncertainties such as variations in incoming component positions or changes in connection points, adjusting its movements in real time to ensure accuracy.
When handling flexible wiring harnesses, the robot dynamically modulates its applied force, ensuring secure connections while avoiding damage to sensitive components.
During live production, Xiao Mo has maintained a connection success rate exceeding 99%, with operating speed comparable to that of experienced human workers.
Xiao Mo has now become an integral part of the production line. Beyond executing assembly tasks, it can independently inspect wiring connections, flag abnormalities in real time, and help reduce defect rates. During idle intervals, the robot switches proactively into inspection mode. In continuous production involving multiple battery models, its daily workload has reportedly tripled, while delivering high levels of consistency and operational stability.
Xiao Mo was developed by Spirit AI, a company within CATL's broader ecosystem, and is powered by CATL's in-house battery technology, reflecting close collaboration across the industrial value chain. To translate real-world production demands into concrete engineering specifications, CATL coordinated cross-departmental research directly on the factory floor, jointly formulating a development plan designed to be both forward-looking and practically deployable. This groundwork has paved the way for large-scale deployment of embodied intelligent robots.
Looking ahead, CATL plans to build on the successful rollout of Xiao Mo to further raise the level of automation and intelligence across its PACK production lines. The company aims to deepen the integration between smart manufacturing and green energy technologies, while accelerating the application of embodied intelligence models across a broader range of scenarios, in support of global zero-carbon ambitions.
