Research
ETH Zürich Teaches a Humanoid to Climb, Vault and Recover by Planning Body Contacts
ETH Zürich has developed a humanoid capable of climbing, vaulting, and recovering through advanced planning techniques.
Researchers at ETH Zürich have successfully taught a humanoid robot to perform complex mobility tasks such as climbing and vaulting. This was achieved by integrating human motion data with reinforcement learning to enhance the robot's ability to plan body contacts during movement.
AISummary written by AI from the source below. Check the originals for details.
Key facts
- Research conducted at ETH Zürich
- Humanoid tested on a Unitree G1
- Combines human motion data with reinforcement learning
Sources · 1
Humanoids Daily ·
ETH Zürich Teaches a Humanoid to Climb, Vault and Recover by Planning Body Contacts
Tested on a Unitree G1, the research combines human motion data and reinforcement learning to connect mobility skills using onboard…
Open source