aiqbench analysis
RoboRecover introduces a critical benchmark for evaluating robot policies under execution deviations, addressing the gap in current assessments that often ignore dynamic interaction failures during closed-loop operation. By reconstructing off-nominal intermediate states from action prefixes, this framework enables researchers to measure how well agents recover task progress after unexpected contacts or object relation changes. This work establishes recovery capability as a distinct and necessary dimension for advancing robust autonomous robotics systems.
RoboRecover: Benchmarking Robot Policy Recovery under Execution Deviations
arXiv:2609.28952v1 Announce Type: new Abstract: Robot-policy benchmarks increasingly cover diverse tasks and preset out-of-distribution conditions, but typically evaluate complete trajectories from predefined initial states. These evaluations often focus on the initialized scene and the final outcome, while paying less attention to the dynamic interaction process. During closed-loop execution, actions and contacts can alter object relations and task progress, producing off-nominal intermediate states that need recovery. Recovery requires a policy to infer how task progress has changed, correct the relevant relations, and continue the original goal. We introduce RoboRecover, a benchmark for robot policy recovery under execution deviations. RoboRecover selects deviation states from trajectories, reconstructs them by replaying action prefixes, and evaluates policies on the original task. RoboRecover contains 2,000 scenarios across RoboTwin and LIBERO, with 1,000 scenarios and a fixed 800/200 train/test split on each platform. Results show that initial-state performance does not determine recovery performance and policies exhibit different recovery strengths across scenarios. Using its training split, RoboRecover further supports study on recovery interventions. RoboRecover establishes recovery from execution-induced intermediate states as a distinct dimension of robot policy evaluation.
Abstract:Robot-policy benchmarks increasingly cover diverse tasks and preset out-of-distribution conditions, but typically evaluate complete trajectories from predefined initial states. These evaluations often focus on the initialized scene and the final outcome, while paying less attention to the dynamic interaction process. During closed-loop execution, actions and contacts can alter object relations and task progress, producing off-nominal intermediate states that need recovery. Recovery requires a policy to infer how task progress has changed, correct the relevant relations, and continue the original goal. We introduce RoboRecover, a benchmark for robot policy recovery under execution deviations. RoboRecover selects deviation states from trajectories, reconstructs them by replaying action prefixes, and evaluates policies on the original task. RoboRecover contains 2,000 scenarios across RoboTwin and LIBERO, with 1,000 scenarios and a fixed 800/200 train/test split on each platform. Results show that initial-state performance does not determine recovery performance and policies exhibit different recovery strengths across scenarios. Using its training split, RoboRecover further supports study on recovery interventions. RoboRecover establishes recovery from execution-induced intermediate states as a distinct dimension of robot policy evaluation.
Robotics (cs.RO)
arXiv:2609.28952 [cs.RO]
(or arXiv:2609.28952v1 [cs.RO] for this version)
https://doi.org/10.48550/arXiv.2609.28952 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 268 words
Related coverage
CARE: Experience-Guided Atomic Corrective Execution for Vision-Language-Action Policies
arXiv · Sep 22, 2026
RESample: A Robust Data Augmentation Framework via Exploratory Sampling for Robotic Manipulation
arXiv · Aug 6, 2026
Benchmarking Vision-Language-Action Models on SO-101: Failure and Recovery Analysis
arXiv · Jun 16, 2026
Benchmarking Vision-Language-Action Models on SO-101: Failure and Recovery Analysis
arXiv · Jun 9, 2026