Benchmarking Generalizable Bimanual Manipulation: RoboTwin Dual-Arm Collaboration Challenge at CVPR 2025 MEIS Workshop
Abstract
Embodied Artificial Intelligence (Embodied AI) is an emerging frontier in robotics, driven by the need for autonomous systems that can perceive, reason, and act in complex physical environments. While single-arm systems have shown strong task performance, collaborative dual-arm systems are essential for handling more intricate tasks involving rigid, deformable, and tactile-sensitive objects. To advance this goal, we launched the RoboTwin Dual-Arm Collaboration Challenge at the 2nd MEIS Workshop, CVPR 2025. Built on the RoboTwin Simulation platform (1.0 and 2.0) and the AgileX COBOT-Magic Robot platform, the competition consisted of three stages: Simulation Round 1, Simulation Round 2, and a final Real-World Round. Participants totally tackled 17 dual-arm manipulation tasks, covering rigid, deformable, and tactile-based scenarios. The challenge attracted 64 global teams and over 400 participants, producing top-performing solutions like SEM and AnchorDP3 and generating valuable insights into generalizable bimanual policy learning. This report outlines the competition setup, task design, evaluation methodology, key findings and future direction, aiming to support future research on robust and generalizable bimanual manipulation policies. The Challenge Webpage is available at https://robotwin-benchmark.github.io/cvpr-2025-challenge/.
Keywords
Cite
@article{arxiv.2506.23351,
title = {Benchmarking Generalizable Bimanual Manipulation: RoboTwin Dual-Arm Collaboration Challenge at CVPR 2025 MEIS Workshop},
author = {Tianxing Chen and Kaixuan Wang and Zhaohui Yang and Yuhao Zhang and Zanxin Chen and Baijun Chen and Wanxi Dong and Ziyuan Liu and Dong Chen and Tianshuo Yang and Haibao Yu and Xiaokang Yang and Yusen Qin and Zhiqiang Xie and Yao Mu and Ping Luo and Tian Nian and Weiliang Deng and Yiheng Ge and Yibin Liu and Zixuan Li and Dehui Wang and Zhixuan Liang and Haohui Xie and Rijie Zeng and Yunfei Ge and Peiqing Cong and Guannan He and Zhaoming Han and Ruocheng Yin and Jingxiang Guo and Lunkai Lin and Tianling Xu and Hongzhe Bi and Xuewu Lin and Tianwei Lin and Shujie Luo and Keyu Li and Ziyan Zhao and Ke Fan and Heyang Xu and Bo Peng and Wenlong Gao and Dongjiang Li and Feng Jin and Hui Shen and Jinming Li and Chaowei Cui and Yu Chen and Yaxin Peng and Lingdong Zeng and Wenlong Dong and Tengfei Li and Weijie Ke and Jun Chen and Erdemt Bao and Tian Lan and Tenglong Liu and Jin Yang and Huiping Zhuang and Baozhi Jia and Shuai Zhang and Zhengfeng Zou and Fangheng Guan and Tianyi Jia and Ke Zhou and Hongjiu Zhang and Yating Han and Cheng Fang and Yixian Zou and Chongyang Xu and Qinglun Zhang and Shen Cheng and Xiaohe Wang and Ping Tan and Haoqiang Fan and Shuaicheng Liu and Jiaheng Chen and Chuxuan Huang and Chengliang Lin and Kaijun Luo and Boyu Yue and Yi Liu and Jinyu Chen and Zichang Tan and Liming Deng and Shuo Xu and Zijian Cai and Shilong Yin and Hao Wang and Hongshan Liu and Tianyang Li and Long Shi and Ran Xu and Huilin Xu and Zhengquan Zhang and Congsheng Xu and Jinchang Yang and Feng Xu},
journal= {arXiv preprint arXiv:2506.23351},
year = {2025}
}
Comments
Challenge Webpage: https://robotwin-benchmark.github.io/cvpr-2025-challenge/