Critical Interval MSE: Toward Reliable Offline Validation for Robot Manipulation Policies

2026-06-29

AI Summary

This work addresses the weak correlation between conventional offline evaluation metrics and actual robotic policy performance, which hinders efficient policy iteration. To bridge this gap, the authors propose Critical Interval Mean Squared Error (CI-MSE), a novel offline evaluation method that focuses error assessment on task-critical time intervals and incorporates a lightweight action alignment mechanism to better reflect real-world deployment outcomes. Evaluated across multiple policy checkpoints, CI-MSE achieves a Spearman correlation coefficient of −0.87 with true roll-out performance—significantly outperforming standard MSE (−0.61)—and demonstrates strong robustness to hyperparameter variations. These results indicate that CI-MSE substantially enhances the reliability and practical utility of offline policy validation in robotics.

Abstract

Real-world evaluation is the gold standard for robot policies because it tests them against the physical conditions and deployment challenges they are ultimately designed to handle. However, real-world evaluation is also the bottleneck for iterating on robot policies: it is costly, difficult to reproduce, and often too sparse to reliably compare nearby model variants. A straightforward proxy for performance is validation loss on expert demonstrations, but this proxy is often poorly correlated with real-world performance. In this paper, we introduce Critical Interval MSE (CI-MSE), an intuitively simple yet effective offline validation metric. CI-MSE restricts error computation to task-critical segments and pairs it with simple action-alignment procedures that better match rollout-time behavior. Across simulation and real-world experiments, CI-MSE yields a stronger correlation between validation error and rollout performance than raw MSE. Across a wide range of policy checkpoints, CI-MSE achieves a Spearman's rank correlation of $-0.87$, much closer to the ideal value of $-1$ than raw MSE's $-0.61$, demonstrating a significant improvement. We show through sensitivity analysis that our metric is robust to a wide range of hyperparameters. We further study the effectiveness of CI-MSE under evaluation distribution shifts and suggest design boundaries when using this metric. In summary, this paper provides a simple and reliable offline validation tool for accelerating policy iteration. Project webpage: https://ci-mse.github.io/

Problem

• offline validation
• robot manipulation
• policy evaluation
• performance correlation
• validation metric

Innovation

• Critical Interval MSE
• offline validation
• robot manipulation
• policy evaluation
• action alignment