WorldSample: Closed-loop Real-robot RL with World Modelling
2026-07-02 • Robotics
RoboticsArtificial Intelligence
AI summaryⓘ
The authors developed WorldSample, a method to help robots learn better by combining real experiences with computer-generated ones. They use real robot actions to create a detailed model of the world, producing synthetic examples for training without needing many costly physical trials. Their approach also carefully chooses which synthetic data to use, avoiding confusion from inaccurate simulations. Tests showed their method improved task success and reduced training time compared to other approaches, while also producing clearer simulated visuals.
Reinforcement LearningImitation LearningWorld ModelData AugmentationRobot ManipulationPolicy LearningPhysical RolloutsSample SelectionVisual Fidelity
Authors
Yuquan Xue, Le Xu, Zeyi Liu, Zhenyu Wu, Zhengyi Gu, Xinyang Song, Bofang Jia, Ziwei Wang
Abstract
Reinforcement learning (RL) can overcome the demonstration-coverage limitation of imitation learning (IL) by allowing robots to improve through trial-and-error interaction beyond the states observed in demonstrations. However, deploying RL on real robots remains constrained by high interaction costs, since each physical rollout is costly and reflects only one realized action-outcome path. To address this challenge, we propose WorldSample, a physically grounded data augmentation framework for real-robot RL that closes a real-synthetic loop between physical rollouts, world-model generation, and policy improvement. Grounded on real rollouts, WorldSample generates high-fidelity synthetic transitions through a post-trained world model, which greatly lowers the visual hallucination. Specifically, rather than simply using these transitions as real-world experience, WorldSample introduces Policy-Paced Learning (PPL) to regulate the training process through sample selection and scheduling, balancing useful augmentation against value overestimation and mitigating the hallucination-induced noise. Experiments on robot manipulation tasks involving contact-rich and precise tasks show that WorldSample improves policy success rate by 28% while reducing training steps by 59% compared with baselines. Furthermore, WorldSample improves world model visual fidelity by 19.4dB in PSNR and 0.47 in SSIM over demonstration-only post-training, validating the effectiveness of the real-synthetic loop for both policy and world model performance.