遇见数据集

hello3x3/zte_embodied_2026

收藏
Hugging Face2026-05-26 更新2026-05-31 收录
官方服务:

资源简介:

该数据集是从2026年第十六届中兴捧月全球精英挑战赛 - 神算师算法精英挑战赛 - 具身智能(初赛)的原始数据转换而来,采用LeRobot v3.0格式。数据集包含350个episodes,总计51,467帧,涵盖15个不同的任务,涉及双灵巧臂机器人(双7自由度手臂和双6自由度手)执行抓取和放置操作,例如拾取可乐、绿茶、雪碧、矿泉水、芬达、苹果、橙子、香蕉、玩偶、球和魔方等物体,并将其放入盒子或放置在桌子上。数据以30 FPS的视频(分辨率为1280x720,编码为av1)和26维的关节位置状态及动作命令形式提供,适用于具身智能和机器人学习研究。

This dataset is converted from the raw data of the 2026 16th ZTE Cup Global Elite Challenge - Algorithm Elite Challenge - Embodied Intelligence (Preliminary). It is in LeRobot v3.0 format and includes 350 episodes with a total of 51,467 frames across 15 tasks. The dataset features a dual-arm dexterous robot (dual 7-DOF arms and dual 6-DOF hands) performing pick-and-place operations on objects such as Cocacola, green tea, Sprite, mineral water, Fanta, apple, orange, banana, doll, ball, and Rubiks Cube. Data includes video at 30 FPS (1280x720 resolution, av1 codec) and 26-dimensional joint position states and action commands, suitable for embodied intelligence and robotics learning research.

提供机构:
hello3x3
二维码
社区交流群
二维码
科研交流群
商业服务