In-Context Learning for Robots: Methods and Applications

What happened

arXiv:2609.36012v1 Announce Type: new Abstract: General-purpose robots must infer what a new task requires and translate that understanding into appropriate physical action. In-context learning (ICL) for robots provides support for this process by using demonstrations and interaction to direct existing competence with neural parameters held fixed during deployment.

Comparing these interfaces clarifies their transfer assumptions and the roles of training, correspondence, and memory in making context useful. This analysis links method design to evaluation practices that distinguish responsiveness to teaching, physical transfer, and benefits from retained experience. The resulting agenda connects compositional task acquisition and faithful transfer with physical recursive self-improvement, in which experience improves the ability to learn subsequent tasks.

Key facts

• In-context learning (ICL) for robots — support: this process by using demonstrations and interaction to direct existing competence with neural parameters held fixed during deployment