paper-with-me

홈 › Papers

Rewind-IL: Online Failure Detection and State Respawning for Imitation Learning

2026-04-17 · Gehan Zheng, Sanjay Seenivasan, Matthew Johnson-Roberson, Weiming Zhi arxiv

Imitation learning has enabled robots to acquire complex visuomotor manipulation skills from demonstrations, but deployment failures remain a major obstacle, especially for long-horizon action-chunked policies. Once execution drifts off the demonstration manifold, these policies often continue producing locally plausible actions without recovering from the failure. Existing runtime monitors either require failure data, over-trigger under benign feature drift, or stop at failure detection without providing a recovery mechanism. We present Rewind-IL, a training-free online safeguard framework for generative action-chunked imitation policies. Rewind-IL combines a zero-shot failure detector based on Temporal Inter-chunk Discrepancy Estimate (TIDE), calibrated with split conformal prediction, with a state-respawning mechanism that returns the robot to a semantically verified safe intermediate state. Offline, a vision-language model identifies recovery checkpoints in demonstrations, and the frozen policy encoder is used to construct a compact checkpoint feature database. Online, Rewind-IL monitors self-consistency in overlapping action chunks, tracks similarity to the checkpoint library, and, upon failure, rewinds execution to the latest verified safe state before restarting inference from a clean policy state. Experiments on real-world and simulated long-horizon manipulation tasks, including transfer to flow-matching action-chunked policies, demonstrate that policy-internal consistency coupled with semantically grounded respawning offers a practical route to improved reliability in imitation learning. Supplemental materials are available at https://sjay05.github.io/rewind-il

📄 PDF Abstract BibTeX arXiv:2604.16683

Code (0)

등록된 구현이 없습니다.

Similar Papers 제목 키워드 기반

See, Plan, Rewind: Progress-Aware Vision-Language-Action Models for Robust Robotic Manipulation

2026-03-10 · Tingjun Dai, Mingfei Han, Tingwen Du, Zhiheng Liu 외 arxiv

Measurement of task progress through explicit, actionable milestones is critical for robust robotic manipulation. This progress awareness enables a model to ground its current task status, anticipate verifiable intermedi…

Comparing Rewinding and Fine-tuning in Neural Network Pruning

2020-03-05 · ICLR 2020 1 · Alex Renda, Jonathan Frankle, Michael Carbin

Many neural network pruning algorithms proceed in three steps: train the network to completion, remove unwanted structure to compress the network, and retrain the remaining structure to recover lost accuracy. The standar…

Network Pruning

Reproducibility Study: Comparing Rewinding and Fine-tuning in Neural Network Pruning

2021-09-20 · Szymon Mikler

Scope of reproducibility: We are reproducing Comparing Rewinding and Fine-tuning in Neural Networks from arXiv:2003.02389. In this work the authors compare three different approaches to retraining neural networks after p…

Network Pruning

FedRewind: Rewinding Continual Model Exchange for Decentralized Federated Learning

2024-11-14 · Luca Palazzo, Matteo Pennisi, Federica Proietto Salanitri, Giovanni Bellitto 외

In this paper, we present FedRewind, a novel approach to decentralized federated learning that leverages model exchange among nodes to address the issue of data distribution shift. Drawing inspiration from continual lear…

Continual LearningFederated Learning

ReWind: Understanding Long Videos with Instructed Learnable Memory

2024-11-23 · CVPR 2025 1 · Anxhelo Diko, Tinghuai Wang, Wassim Swaileh, Shiyan Sun 외

Vision-Language Models (VLMs) are crucial for applications requiring integrated understanding textual and visual information. However, existing VLMs struggle with long videos due to computational inefficiency, memory lim…

Large Language ModelQuestion AnsweringVideo UnderstandingVisual Question Answering+1