FOCUS: First-Order Control Using Optimal Transport Supervision for Imitation from Observations
FOCUS: First-Order Control Using Optimal Transport Supervision for Imitation from Observations
This research presents FOCUS, a method that combines local optimal-transport alignment with short-horizon differentiable actor-critic learning to enable imitation learning from observations without requiring temporal correspondence between state-only demonstrations and expert trajectories.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
Reported numbers
muscle activations
152
View original evidence
ion or teacher forcing. Across contact-rich locomotion tasks in DFlex, including a musculoskeletal humanoid controlled by 152 muscle activations, FOCUS improves sample efficiency and training stability over adversarial imitation-from-observation methods and prior differentiable imitation baselines.
Open source S4
- paper: not_reported
- code: not_reported
- weights: not_reported
- dataset: not_reported
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
ion or teacher forcing. Across contact-rich locomotion tasks in DFlex, including a musculoskeletal humanoid controlled by 152 muscle activations, FOCUS improves sample efficiency and training stability over adversarial imitation-from-observation methods and prior differentiable imitation baselines.
Open source S4
Source：arXiv Robotics — research abstracts · arxiv.org