analiza aiqbench
This paper introduces Reward-DAgger, a framework that leverages general-purpose reward models to gate human intervention in robot learning, significantly improving failure detection without task-specific tuning. By decoupling the gating mechanism from policy internals, it enables robust interactive imitation learning across diverse environments and architectures. The method demonstrates superior efficiency in balancing autonomous success rates with human effort, addressing critical deployment challenges for generalist robotic agents.
Reward-DAgger: Robot-Gated Interactive Imitation Learning with General-Purpose Progress-Based Reward Models
arXiv:2610.04054v1 Announce Type: cross Abstract: Recent advances in robot learning have enabled generalist control policies capable of completing a wide range of tasks. However, their performance degrades when deployed in unseen environments, making it critical to detect failures and teach recovery behaviors. Existing runtime monitoring methods often require task- and policy-specific training or hyperparameter tuning, limiting cross-task deployment and introducing additional overhead during iterative policy updates. We present Reward-DAgger, a robot-gated interactive imitation learning framework that uses dense progress signals from a general-purpose reward model to determine when human intervention is needed. Our approach is agnostic to the underlying policy architecture, requires no access to policy internals, and can be applied across tasks without retuning the gating mechanism. Our results show that Reward-DAgger achieves a better failure-detection accuracy-latency tradeoff than existing runtime monitoring baselines. Across eight simulated and real-world tasks, Reward-DAgger consistently improves the downstream policy's autonomous success rate throughout interactive learning and achieves strong return on human effort, outperforming the baselines in most settings. Importantly, the same gating configuration is used across tasks without task-specific hyperparameter tuning, demonstrating transfer across tasks, environments, and policy architectures. Code and videos are available at https://liralab.usc.edu/reward-dagger.
Abstract:Recent advances in robot learning have enabled generalist control policies capable of completing a wide range of tasks. However, their performance degrades when deployed in unseen environments, making it critical to detect failures and teach recovery behaviors. Existing runtime monitoring methods often require task- and policy-specific training or hyperparameter tuning, limiting cross-task deployment and introducing additional overhead during iterative policy updates. We present Reward-DAgger, a robot-gated interactive imitation learning framework that uses dense progress signals from a general-purpose reward model to determine when human intervention is needed. Our approach is agnostic to the underlying policy architecture, requires no access to policy internals, and can be applied across tasks without retuning the gating mechanism. Our results show that Reward-DAgger achieves a better failure-detection accuracy-latency tradeoff than existing runtime monitoring baselines. Across eight simulated and real-world tasks, Reward-DAgger consistently improves the downstream policy's autonomous success rate throughout interactive learning and achieves strong return on human effort, outperforming the baselines in most settings. Importantly, the same gating configuration is used across tasks without task-specific hyperparameter tuning, demonstrating transfer across tasks, environments, and policy architectures. Code and videos are available at this https URL.
Robotics (cs.RO); Artificial Intelligence (cs.AI)
arXiv:2610.04054 [cs.RO]
(or arXiv:2610.04054v1 [cs.RO] for this version)
https://doi.org/10.48550/arXiv.2610.04054 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.
Text integral sub licența deschisă a publicației · 286 cuvinte
Subiecte conexe
BlenDAgger: Blended Shared Control for Interactive Imitation Learning
arXiv · 30 sept. 2026
Vault: One-Step Latent Generation with Positive-Anchored Rewards for Autonomous Driving
arXiv · 29 sept. 2026
Echoverse: Deep, evolving environments for computer-use agents
Microsoft Research · 30 iul. 2026
EvoHIL: Self-Evolving Reward and Flow-Matched Policy Optimization for Robust Human-in-the-Loop Reinforcement Learning
arXiv · 5 aug. 2026