Chinese researchers have introduced Husky, a framework that trains the Unitree G1 humanoid robot to balance on and ride skateboards. Using large-scale simulation, the system teaches the robot to push, steer, and maintain balance on a moving board, then transfers that skill to the real world.
In tests, the G1 rides across outdoor surfaces, turning and adjusting its posture dynamically as the skateboard moves beneath it. The key challenge is real‑time adaptation to unstable contact: Husky must help the robot handle shifting friction, slopes, and small disturbances without wiping out.
This work could generalize to construction sites, disaster zones, and other uneven terrains, where robots must operate on unpredictable surfaces. Separately, Unitree showcased a “winter challenge” where the G1 walked over 130,000 steps in Altay, Xinjiang, at temperatures near −47°C.
The robot is also programmed to trace an Olympic logo pattern in the snow while wearing a puffer jacket for fun. The robot was able to complete the route, demonstrating its durability in extreme cold. Together, these demos highlight how quickly humanoid platforms are improving in both agility and environmental robustness.
Comments
No comments yet. Be the first to share your thoughts!