Beyond Success: Refining Elegant Robot Manipulation from Mixed-Quality Data via Just-in-Time Intervention

TL;DR

This paper introduces a framework for improving robotic manipulation execution quality by using an Elegance Critic and Just-in-Time Intervention, which refine actions based on implicit task constraints without retraining the base policy.

Contribution

It proposes a decoupled refinement framework with an Elegance Critic trained via offline Calibrated Q-Learning and a JITI mechanism for on-demand intervention, enhancing execution quality in VLA models.

Findings

Significant improvement in execution quality on LIBERO-Elegant benchmark.

Effective refinement on unseen manipulation tasks.

JITI mechanism reduces unnecessary interventions.

Abstract

Vision-Language-Action (VLA) models have enabled notable progress in general-purpose robotic manipulation, yet their learned policies often exhibit variable execution quality. We attribute this variability to the mixed-quality nature of human demonstrations, where the implicit principles that govern how actions should be carried out are only partially satisfied. To address this challenge, we introduce the LIBERO-Elegant benchmark with explicit criteria for evaluating execution quality. Using these criteria, we develop a decoupled refinement framework that improves execution quality without modifying or retraining the base VLA policy. We formalize Elegant Execution as the satisfaction of Implicit Task Constraints (ITCs) and train an Elegance Critic via offline Calibrated Q-Learning to estimate the expected quality of candidate actions. At inference time, a Just-in-Time Intervention (JITI) mechanism monitors critic confidence and intervenes only at decision-critical moments, providing selective, on-demand refinement. Experiments on LIBERO-Elegant and real-world manipulation tasks show that the learned Elegance Critic substantially improves execution quality, even on unseen tasks. The proposed model enables robotic control that values not only whether tasks succeed, but also how they are performed.