ComSim: Building Scalable Real-World Robot Data Generation via Compositional Simulation
Compositional simulation: classical sim produces accurate action-video pairs and a neural simulator restyles them as real-looking video, in a real-sim-real loop seeded by a little real data.组合式仿真：经典仿真生成准确的动作-视频对，再由神经模拟器转换成接近真实的画面，形成以少量真机数据为起点的“真-仿-真”数据扩增闭环。
BibTeX
@misc{qin2026comsim,
title = {{ComSim: Building Scalable Real-World Robot Data Generation via Compositional Simulation}},
author = {Yiran Qin and Jiahua Ma and Li Kang and Wenzhan Li and Yihang Jiao and Xin Wen and Xiufeng Song and Heng Zhou and Jiwen Yu and Zhenfei Yin and Xihui Liu and Philip Torr and Yilun Du and Ruimao Zhang},
year = {2026},
eprint = {2604.11386},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2604.11386},
}