Sim-to-Real of Humanoid Locomotion Policies via Joint Torque Space Perturbation Injection

1. 1Inject state-dependent perturbations into the input joint torque during forward simulation
2. 2Design perturbations to simulate a broader spectrum of reality gaps than standard domain randomization
3. 3Train control policies using simulated experiences with perturbation injection
4. 4Evaluate and refine policies in real-world scenarios

Robotics and AI engineers on a team can benefit from this research as it provides a novel approach to sim-to-real transfer, allowing for more robust and realistic control policies. This can be particularly useful in humanoid robotics applications where sim-to-real transfer is crucial

💡 Injecting state-dependent perturbations into joint torque space can improve sim-to-real transfer for humanoid locomotion policies

🤖 Novel sim-to-real method for humanoid locomotion policies using joint torque perturbations!