Task-Specialized Language-Conditioned Models for Fine-Grained Robotic Manipulation

arXiv:2611.00955 · cs.RO · Submitted 2026-10-31 · Read on arXiv

cs.RO

Submitted: 2026-10-31

Updated: 2026-10-31

Comments: 5 pages, 2 figures, 2 tables. Technical report for the MetaFine Open Competition at CoRL 2026

License: http://creativecommons.org/licenses/by/4.0/

The gist: Fine-grained robotic manipulation requires instruction grounding, robust perception, and precise contact control.

Terminology

Abstract

Fine-grained robotic manipulation requires instruction grounding, robust perception, and precise contact control. We describe our submission to the MetaFine Open Competition at CoRL 2026. Five independently trained models use a common architecture with frozen DINOv2 and CLIP encoders, trainable language-conditioned feature modulation, and task-specific action heads. We improve demonstrations through contact-aware grasp filtering, instruction-matched data, and expert correction of insertion failures. At inference, task-dependent strategies include photometric test-time augmentation, stochastic action averaging, and uncertainty-triggered replanning. Our submission achieves an official overall score of 0.914 and ranks third among 15 teams. We analyze the models and their task-specific behavior.

Sources

Related papers