Saved in:
Bibliographic Details
Main Authors: Snyder, David, Badithela, Apurva, Matni, Nikolai, Pappas, George, Majumdar, Anirudha, Itkina, Masha, Nishimura, Haruki
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2603.13616
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912965910855680
author Snyder, David
Badithela, Apurva
Matni, Nikolai
Pappas, George
Majumdar, Anirudha
Itkina, Masha
Nishimura, Haruki
author_facet Snyder, David
Badithela, Apurva
Matni, Nikolai
Pappas, George
Majumdar, Anirudha
Itkina, Masha
Nishimura, Haruki
contents Generalist robot manipulation policies are becoming increasingly capable, but are limited in evaluation to a small number of hardware rollouts. This strong resource constraint in real-world testing necessitates both more informative performance measures and reliable and efficient evaluation procedures to properly assess model capabilities and benchmark progress in the field. This work presents a novel framework for robot policy comparison that is sample-efficient, statistically rigorous, and applicable to a broad set of evaluation metrics used in practice. Based on safe, anytime-valid inference (SAVI), our test procedure is sequential, allowing the evaluator to stop early when sufficient statistical evidence has accumulated to reach a decision at a pre-specified level of confidence. Unlike previous work developed for binary success, our unified approach addresses a wide range of informative metrics: from discrete partial credit task progress to continuous measures of episodic reward or trajectory smoothness, spanning both parametric and nonparametric comparison problems. Through extensive validation on simulated and real-world evaluation data, we demonstrate up to 70% reduction in evaluation burden compared to standard batch methods and up to 50% reduction compared to state-of-the-art sequential procedures designed for binary outcomes, with no loss of statistical rigor. Notably, our empirical results show that competing policies can be separated more quickly when using fine-grained task progress than binary success metrics.
format Preprint
id arxiv_https___arxiv_org_abs_2603_13616
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Beyond Binary Success: Sample-Efficient and Statistically Rigorous Robot Policy Comparison
Snyder, David
Badithela, Apurva
Matni, Nikolai
Pappas, George
Majumdar, Anirudha
Itkina, Masha
Nishimura, Haruki
Robotics
Applications
Generalist robot manipulation policies are becoming increasingly capable, but are limited in evaluation to a small number of hardware rollouts. This strong resource constraint in real-world testing necessitates both more informative performance measures and reliable and efficient evaluation procedures to properly assess model capabilities and benchmark progress in the field. This work presents a novel framework for robot policy comparison that is sample-efficient, statistically rigorous, and applicable to a broad set of evaluation metrics used in practice. Based on safe, anytime-valid inference (SAVI), our test procedure is sequential, allowing the evaluator to stop early when sufficient statistical evidence has accumulated to reach a decision at a pre-specified level of confidence. Unlike previous work developed for binary success, our unified approach addresses a wide range of informative metrics: from discrete partial credit task progress to continuous measures of episodic reward or trajectory smoothness, spanning both parametric and nonparametric comparison problems. Through extensive validation on simulated and real-world evaluation data, we demonstrate up to 70% reduction in evaluation burden compared to standard batch methods and up to 50% reduction compared to state-of-the-art sequential procedures designed for binary outcomes, with no loss of statistical rigor. Notably, our empirical results show that competing policies can be separated more quickly when using fine-grained task progress than binary success metrics.
title Beyond Binary Success: Sample-Efficient and Statistically Rigorous Robot Policy Comparison
topic Robotics
Applications
url https://arxiv.org/abs/2603.13616