China’s Unitree Unveils Robot Dog That Sprints at 11 mph, Lifts 143 Pounds
Chinese robotics firm Unitree Robotics has introduced the As2, a new quadruped robot designed to blend industrial-grade strength with advanced AI.
The company says the machine is built for real-world environments, offering a mix of speed, payload capacity, and autonomous features. The As2 arrives as the latest entry in Unitree’s growing lineup of robot dogs, but this one pushes performance further with higher torque, longer runtime, and expanded developer tools.
Weighing about 18 kg with its battery, the As2 delivers up to 90 N·m of maximum joint torque in its highest configuration. That power allows the robot to reach speeds of more than 5 m/s (around 11 mph) under no-load conditions.
Unitree says the robot’s torque-to-weight ratio stands at roughly 5 N·m/kg, giving it the strength to move quickly while maintaining balance. The quadruped can climb stairs up to 25 cm high, handle slopes of about 40 degrees, and even mount vertical platforms as tall as 50 cm, depending on the version.
The As2 comes in three configurations (Air, Pro, and Edu) with varying performance levels. The Edu model offers the highest torque and speed, though peak speed can only be sustained briefly.
Beyond speed, payload is a key selling point. When standing still, the Edu variant can support up to 65 kg. During continuous walking, the maximum load drops to about 15 kg, depending on the model. Lower-tier versions carry less, but still offer notable capacity for a machine of this size.
Battery life also differs by configuration. The Pro and Edu models use a 15,000mAh (648Wh) battery, offering more than four hours of runtime when unloaded and a walking range exceeding 20 kilometers in laboratory conditions. With load, the runtime falls to around 2.5 hours and roughly 13 kilometers of range. The Air model uses a smaller 8,000mAh battery with shorter endurance.
Designed for harsh conditions
The As2 is rated IP54, meaning it is resistant to dust and rain. It is built to operate in temperatures ranging from -20°C to 50°C, making it suitable for outdoor and industrial settings.
The robot’s frame combines aluminum and high-strength plastic. Each leg features multiple degrees of freedom, powered by low-inertia permanent magnet synchronous motors. Industrial-grade bearings are integrated for precision and load capacity.
AI brain and smart following
At its core, the As2 runs on an 8-core CPU. Higher-end versions include the company’s ISS 3.0 Intelligent Follow System, which enables centimeter-level positioning for side-following and tracking.
The Pro and Edu variants also integrate industrial-grade LiDAR systems, while the Air version uses Unitree’s L2 ultra-wide-angle LiDAR. GPS and 4G connectivity are included but disabled by default and require user authorization.
For developers, the Edu model supports secondary development and offers optional expansion with an Nvidia Jetson Orin NX module. Unitree says the open platform allows support for large AI models, enabling embodied AI interaction and autonomous decision-making.
Connectivity options include Wi-Fi 6, Bluetooth 5.2, Gigabit Ethernet, SBUS, along with a speaker, microphone, HD camera, and front light across all versions.
Availability and pricing
Unitree has not publicly disclosed pricing for any As2 variant. All models are listed as “Contact Sales” on the company’s website. Warranty coverage spans six months for the Air model and 12 months for Pro and Edu versions. A charging dock is available for the Edu configuration.
The company notes in its documentation that data “was obtained under laboratory conditions,” adding “actual performance may vary based on usage conditions.” Unitree also cautions that “the quadruped robot has a complex structure and powerful dynamics,” advising users to keep a safe distance during operation.
Also read: Unitree’s ambitions extend past robot dogs, with humanoid robot shipments projected to reach up to 20,000 units in 2026.
The post China’s Unitree Unveils Robot Dog That Sprints at 11 mph, Lifts 143 Pounds appeared first on eWEEK.
