Scaling Bimanual Household Manipulation from 1,500 hours of Demonstrations to On-Policy Corrections

• URL: http://arxiv.org/abs/2609.03591v1
• Date: Thu, 03 Sep 2026 09:37:09 GMT
• Title: Scaling Bimanual Household Manipulation from 1,500 hours of Demonstrations to On-Policy Corrections
• Abstract summary: We release 1,500 hours of diverse bimanual manipulation demonstrations covering everyday household tasks.<n>We use this comprehensive corpus to train XR-2, a powerful vision-language-action (VLA) model.<n>We study two critical scaling axes: varying the amount of expert demonstration data, and post training on DAgger correction data from real time human interventions.
• Score: 20.699475306823462
• License: http://creativecommons.org/licenses/by-sa/4.0/
• Abstract: Learning generalist policies for robust bimanual manipulation is bottlenecked by the scarcity of high quality large scale human demonstration data. In this work, we release 1,500 hours of diverse bimanual manipulation demonstrations covering everyday household tasks, and use this comprehensive corpus to train XR-2, a powerful vision-language-action (VLA) model. Enabled by a purpose built high throughput data pipeline and a carefully designed multi stage training paradigm, XR-2 attains strong manipulation performance in our systematic experiments while retaining favorable training efficiency and high data utilization. We further study two critical scaling axes: varying the amount of expert demonstration data, and post training on DAgger correction data from real time human interventions. In both settings, task success rate improves steadily over the data ranges we probe, exhibiting a clear consistent scaling trend at our current data scale. These results validate both the learning capacity of XR-2 and the promising scaling properties of the released dataset, which we open source to support reproducible research on bimanual robot manipulation learning.

Related papers

• Representation-Conditioned Diffusion Models for Guided Training Data Generation [1.9517610560768623]
  We evaluate the classification performance of models trained on synthetic image datasets produced by generative deep learning.<n>We use latent diffusion models conditioned on learned representations from DINOv2, DINOv3, and CLIP.<n>Our results demonstrate that this representation-conditioned formulation significantly outperforms class-conditioned generation by a large margin.
  arXiv  Detail & Related papers  (2026-05-26T17:32:50Z)
• Towards High Data Efficiency in Reinforcement Learning with Verifiable Reward [54.708851958671794]
  We propose a Data-Efficient Policy Optimization pipeline that combines optimized strategies for both offline and online data selection.<n>In offline phase, we curate a high-quality subset of training samples based on diversity, influence, and appropriate difficulty.<n>During online RLVR training, we introduce a sample-level explorability metric to dynamically filter samples with low exploration potential.
  arXiv  Detail & Related papers  (2025-09-01T10:04:20Z)
• Is Diversity All You Need for Scalable Robotic Manipulation? [50.747150672933316]
  We investigate the nuanced role of data diversity in robot learning by examining three critical dimensions-task (what to do), embodiment (which robot to use), and expert (who demonstrates)-challenging the conventional intuition of "more diverse is better"<n>We show that task diversity proves more critical than per-task demonstration quantity, benefiting transfer from diverse pre-training tasks to novel downstream scenarios.<n>We propose a distribution debiasing method to mitigate velocity ambiguity, the yielding GO-1-Pro achieves substantial performance gains of 15%, equivalent to using 2.5 times pre-training data.
  arXiv  Detail & Related papers  (2025-07-08T17:52:44Z)
• LEAD: Iterative Data Selection for Efficient LLM Instruction Tuning [22.242445543184264]
  We propose LEAD, an efficient iterative data selection framework that accurately estimates sample utility entirely within the standard training loop.<n>Experiments show that LEAD significantly outperforms state-of-the-art methods, improving average model performance by 6.1%-10.8% while using only 2.5% of the training data and reducing overall training time by 5-10x.
  arXiv  Detail & Related papers  (2025-05-12T10:57:51Z)
• How to Achieve Higher Accuracy with Less Training Points? [2.1834099301440526]
  We propose a technique based on influence functions to determine which training samples should be included in the training set.<n>Our approach demonstrates performance comparable to that of training on the entire dataset while using only 10% of the data.
  arXiv  Detail & Related papers  (2025-04-18T09:38:26Z)
• AgiBot World Colosseo: A Large-scale Manipulation Platform for Scalable and Intelligent Embodied Systems [88.05152114775498]
  AgiBot World is a large-scale platform comprising over 1 million trajectories across 217 tasks in five deployment scenarios.<n>AgiBot World guarantees high-quality and diverse data distribution.<n>GO-1 exhibits exceptional capability in real-world dexterous and long-horizon tasks.
  arXiv  Detail & Related papers  (2025-03-09T15:40:29Z)
• Efficient Grammatical Error Correction Via Multi-Task Training and Optimized Training Schedule [55.08778142798106]
  We propose auxiliary tasks that exploit the alignment between the original and corrected sentences. We formulate each task as a sequence-to-sequence problem and perform multi-task training. We find that the order of datasets used for training and even individual instances within a dataset may have important effects on the final performance.
  arXiv  Detail & Related papers  (2023-11-20T14:50:12Z)
• Multi-dataset Training of Transformers for Robust Action Recognition [75.5695991766902]
  We study the task of robust feature representations, aiming to generalize well on multiple datasets for action recognition. Here, we propose a novel multi-dataset training paradigm, MultiTrain, with the design of two new loss terms, namely informative loss and projection loss. We verify the effectiveness of our method on five challenging datasets, Kinetics-400, Kinetics-700, Moments-in-Time, Activitynet and Something-something-v2.
  arXiv  Detail & Related papers  (2022-09-26T01:30:43Z)
• Fix your Models by Fixing your Datasets [0.6058427379240697]
  Current machine learning tools lack streamlined processes for improving the data quality. We introduce a systematic framework for finding noisy or mislabelled samples in the dataset. We demonstrate the efficacy of our framework on public as well as private enterprise datasets of two Fortune 500 companies.
  arXiv  Detail & Related papers  (2021-12-15T02:41:50Z)
• A Framework for Efficient Robotic Manipulation [79.10407063260473]
  We show that a single robotic arm can learn sparse-reward manipulation policies from pixels. We show that, given only 10 demonstrations, a single robotic arm can learn sparse-reward manipulation policies from pixels.
  arXiv  Detail & Related papers  (2020-12-14T22:18:39Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.