CLAM: Continuous Latent Action Models for Robot Learning from Unlabeled Demonstrations
Classification
View four quadrants- Major category
- Related resources
- Quadrant
- Not applicable
- Architecture
- Not applicable
- Prediction paradigm
- Not applicable
- Subcategories
- Latent-action & representation methods
- Source review status
- Not assigned
Category review. Continuous latent inverse/forward modeling provides pseudo-action labels and a grounded decoder for learning a separate visuomotor policy from demonstrations. It is a relevant action-representation method; the forward model is discarded for deployment and no language-conditioned policy is specified. The paper is retained for its adjacent resource/method role; WAM joint-prediction/IDM architecture quadrants do not apply to the cataloged contribution. Reading evidence
Contribution
CLAM learns continuous action codes from robot observation transitions, grounds them with limited action-labeled data, then imitates expert videos in that latent space. Deployment uses a policy and action decoder; future-observation prediction serves training. Its strongest evidence concerns action-label scarcity within one robot embodiment, with several unresolved protocol details.
Abstract
An abstract has not been added yet.
Affiliations
Thomas Lord Department of Computer Science, University of Southern California; Ming Hsieh Department of Electrical and Computer Engineering, University of Southern California