Warner
Cofounder · Robot Learning Research
Warner's background has been research heavy, but it has been hands on with real systems. At Shanghai AI Lab, Warner explored differentiable simulation, trained a Unitree H1-2 humanoid to kneel and walk, and built reinforcement learning motion retargeting. At ShanghaiTech, Warner trained generative models for real time humanoid motion and implemented ROS2 retargeting and tracking. He also implemented and debugged planning and control on a real autonomous race car.
Every real world failure should become a durable evaluation that makes the next policy better.
Expertise
- Robot policy learning
- Humanoid robotics
- World models and simulation
- Robotics data quality