RoCo Challenge Dataset
收藏资源简介:
RoCo Challenge Dataset是由南洋理工大学等机构联合创建的高精度行星齿轮箱装配数据集,旨在推动工业机器人协作操纵研究。数据集包含仿真和现实世界两个版本,每个版本均提供300余条多模态演示数据,涵盖RGB图像、深度信息及机器人本体感知状态。数据通过Isaac Sim仿真环境和Galaxea R1 Lite双臂机器人平台采集,包含从零装配、部分状态恢复及错误修正三类任务场景。该数据集特别关注工业装配中的毫米级精度要求与协作特性,为开发长时序多任务学习算法提供了基准平台,可应用于智能制造、自主机器人等领域的算法验证与系统开发。
RoCo Challenge Dataset is a high-precision planetary gearbox assembly dataset jointly developed by Nanyang Technological University and other institutions, aiming to advance research on industrial robot collaborative manipulation. The dataset includes two versions: simulation and real-world, each providing over 300 multimodal demonstration samples covering RGB images, depth data, and robot proprioceptive sensing states. Data was collected using the Isaac Sim simulation environment and the Galaxea R1 Lite dual-arm robotic platform, encompassing three task scenarios: assembly from scratch, partial state recovery, and error correction. This dataset specifically focuses on the millimeter-level precision requirements and collaborative characteristics in industrial assembly, serving as a benchmark platform for developing long-sequence multi-task learning algorithms, and can be applied to algorithm verification and system development in fields such as intelligent manufacturing and autonomous robotics.

- 1RoCo Challenge at AAAI 2026: Benchmarking Robotic Collaborative Manipulation for Assembly Towards Industrial Automation南洋理工大学; 北京邮电大学; 卡内基梅隆大学; 新加坡科技研究局 · 2026年



