Unitree wants its humanoid robots to learn to improve themselves with AI
Unitree Robotics founder Wang Xingxing presented at the company's physical AI conference in Hangzhou a "self-evolving" robot development system called "Unitree Rehearsal", in which AI agents search research papers, write control code, test it in simulation and verify the best results for use on real robots. Wang told the South China Morning Post he expects AI models to take on more planning and iteration. The company has tested OpenAI's GPT-6 on its robots.
In 40 seconds
- Unitree presented "Unitree Rehearsal", a system that seeks to automate the training cycle of its humanoids with AI agents
- The loop includes searching research papers, writing control code, testing in simulation and verifying results on real robots
- Wang argues that stronger base AI models and more robots in the real world would generate more data and speed up skill acquisition
IASummary generated by artificial intelligence from a story by Xataka. It may contain errors: the original source prevails.




Comments
Be respectful. Comments reported by several people are hidden until a moderator reviews them.
You do not need an account to read, only to comment. Sign in with a code sent to your email or with your Google account. We only show the name you choose, never your email.