Visual-Geometric Policy Steering from In-Context Aligned Abstractions
Authors:
Arthur Fender Coelho Bucker1, Pramod Anantharam 1Jonathan Francis1,2, Jean Oh1,3
Affiliations:
1Carnegie Mellon University 2Bosch Center for AI
3Lavoro AI Research
Three-minute overview
Abstract
Adapting robot policies to out-of-distribution (OOD) settings with limited real-world data remains a key bottleneck to generalist robots. The multimodality and scarcity of robotics data lead end-to-end approaches to overfit and break under small distribution shifts. Test-time adaptation can alleviate this by correcting frozen policies at inference, but most methods ground their correction in a signal learned implicitly from robotic datasets, making it susceptible to the same shifts as the policy it corrects. Recent methods that instead exploit explicit spatial information are typically limited to coarse corrections, missing the nuanced, task-relevant motion most manipulation tasks require. We correct existing policies by transferring the nuanced motion of successful in-distribution rollouts, captured along meaningful geometric axes, known here as object-relational distance pairs. Our framework leverages the semantic understanding of Large Vision Models (LVMs) and Vision-Language Models (VLMs) reasoning, which combines the task-relevant entity pairs and their object-relational distance projections into a single tailored visual-geometric guidance signal. In a simulation benchmark, it efficiently corrects flow-matching policies under pose and task-goal shifts; on a real robot experiment, it widens the region of the workspace the frozen policy covers, more than doubling its success rate on cable placements one grid step outside the training positions, and steers a single-task policy toward objects it was never trained on while nearly eliminating object confusion.