Search papers, labs, and topics across Lattice.
CoRMA, a context-based meta-adaptation framework, enhances Robotic Motor Adaptation (RMA) for force-sensitive assembly tasks by replacing direct simulator-parameter adaptation with a learned 6D semantic contact context. A Transformer adapter, trained with semantic regression and a force-regime contrastive objective, infers this context from force, proprioceptive, and action histories during deployment. Experiments on simulated and real robotic assembly tasks demonstrate that CoRMA retains higher real-world success compared to baselines, suggesting the viability of semantic contact inference as a reusable adaptation interface.
Forget brittle sim-to-real transfer; CoRMA learns a compact, reusable 6D "contact context" that lets robots adapt to force-dominant assembly tasks on the fly.
We present CoRMA(Contrastive Robotic Motor Adaptation), a context-based meta-adaptation framework that modifies RMA for force-dominant assembly. CoRMA replaces raw simulator-parameter adaptation with a compact 6D simulator-only semantic contact context describing contact onset, lateral engagement, guided transition, contact direction, and jamming. A deployable causal Transformer adapter infers this context online from force, proprioceptive, and action histories using semantic regression and a force-regime contrastive objective. At deployment, oracle context is removed and replaced by the inferred context, enabling within-episode adaptation without demonstrations, privileged inputs, or gradient updates. We evaluate CoRMA on PegInsert, GearMesh, and NutThread in Isaac Lab / Isaac Sim~5.0 and on a real Marvin arm. Compared with FORGE baselines that achieve high simulation success but degrade substantially on hardware, CoRMA retains higher verified real success under controlled target-pose noise. These results support semantic contact inference as a reusable adaptation interface within a related assembly task family, while broader unseen-task generalization and Real2Sim calibration remain future work.
