RLBench
收藏arXiv2025-09-30 收录
下载链接：
https://github.com/younggyoseo/apv
下载链接
链接失效反馈官方服务：
资源简介：
该数据集包含了从机器人操作任务中收集的视频，用于预训练无需动作输入的视频预测模型。它包含了10个演示，每个演示通过5个不同的相机视角进行渲染，覆盖了99个不同的任务。总规模达到了4950个视频，涵盖了99项任务，这些任务均属于机器人操作类别。
This dataset consists of videos collected from robot manipulation tasks, and is designed for pre-training video prediction models that do not require action inputs. It includes 10 demonstrations for each of the 99 distinct tasks, with each demonstration rendered from 5 different camera viewpoints. The total scale of the dataset is 4950 videos, covering 99 tasks that all fall under the category of robot manipulation.
提供机构：
RLBench
