aiqbench analysis
CORE advances robot imitation learning by leveraging action-free visual demonstrations, such as human videos, to extract stable terminal state configurations rather than relying on costly robotic trajectories. This approach bridges the embodiment gap between humans and robots, significantly improving task success rates across simulated environments like Meta-World and RoboTwin 2.0 compared to text-conditioned methods. By injecting geometric goal prototypes directly into robot policies, the framework offers a scalable path toward utilizing abundant human data for autonomous manipulation tasks.
CORE: Common Outcome Regularities from Action-Free Visual Demonstrations for Robot Manipulation
arXiv:2606.29517v2 Announce Type: replace Abstract: Robot imitation learning often relies on costly robot demonstrations, while abundant action-free visual demonstrations, such as human videos, are difficult to use because they lack robot-executable actions and suffer from embodiment gaps. We propose CORE, a policy learning framework that extracts Common Outcome Regularities (CORE) from visual demonstrations. Rather than transferring explicit actions across embodiments, CORE exploits a key observation: although successful trajectories for the same task can be diverse, their terminal states often share stable object configurations, spatial relations, and contact constraints. CORE first trains a terminal outcome encoder with contrastive and auxiliary temporal objectives, then aggregates successful terminal embeddings into visual goal prototypes, and finally injects these prototypes as global goal conditions into robot policies. Compared with language instructions, visual goal prototypes provide more concrete geometric and physical constraints for task completion. Across Meta-World, RoboTwin 2.0, and real-world manipulation, CORE improves the average success rate of the corresponding policy backbones by up to +3.9, +11.1, and +17.0 percentage points, respectively, and outperforms text-conditioned variants under the evaluated settings. The project and code are available at https://logssim.github.io/CORE.github.io/.
Abstract:Robot imitation learning often relies on costly robot demonstrations, while abundant action-free visual demonstrations, such as human videos, are difficult to use because they lack robot-executable actions and suffer from embodiment gaps. We propose CORE, a policy learning framework that extracts Common Outcome Regularities (CORE) from visual demonstrations. Rather than transferring explicit actions across embodiments, CORE exploits a key observation: although successful trajectories for the same task can be diverse, their terminal states often share stable object configurations, spatial relations, and contact constraints. CORE first trains a terminal outcome encoder with contrastive and auxiliary temporal objectives, then aggregates successful terminal embeddings into visual goal prototypes, and finally injects these prototypes as global goal conditions into robot policies. Compared with language instructions, visual goal prototypes provide more concrete geometric and physical constraints for task completion. Across Meta-World, RoboTwin 2.0, and real-world manipulation, CORE improves the average success rate of the corresponding policy backbones by up to +3.9, +11.1, and +17.0 percentage points, respectively, and outperforms text-conditioned variants under the evaluated settings. The project and code are available at this https URL.
Robotics (cs.RO)
arXiv:2606.29517 [cs.RO]
(or arXiv:2606.29517v2 [cs.RO] for this version)
https://doi.org/10.48550/arXiv.2606.29517 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 271 words
Related coverage
Runway’s WorldPrompt and the Engineering of Real-Time Worlds
Latent Space · Sep 25, 2026
The Entangled Dimensions of Decision Theory
LessWrong · Jul 30, 2026
Game Design Creativity evals
LessWrong · Oct 12, 2026
OpenRouter: from Seed to Stripe — with OpenRouter’s Alex Atallah & AMP’s Anjney Midha
Latent Space · Sep 25, 2026