BrandonAL/eval_smolvla_spline_gpu_c100_libero_goal
收藏官方服务：
资源简介：
这是一个机器人操作数据集，包含10个不同任务，每个任务由多个episode组成，每个episode包含连续的帧数据。每个样本记录了两个摄像头图像（observation.images.image和observation.images.image2）、机器人状态（34维浮点数）、动作（7维浮点数，包括原始动作）、奖励、完成标志和成功标志。该数据集适用于模仿学习、强化学习等机器人控制任务。
This is a robot manipulation dataset containing 10 different tasks, each consisting of multiple episodes with continuous frame data. Each sample records two camera images (observation.images.image and observation.images.image2), robot state (34-dimensional float), action (7-dimensional float, including raw action), reward, done flag, and success flag. The dataset is suitable for imitation learning, reinforcement learning, and other robot control tasks.
提供机构：
BrandonAL
