rslxcvg/banana_act_direct_color_simple_v1

该数据集是使用LeRobot创建的机器人学数据集，主要用于机器人控制任务。数据集包含动作、观测状态和前视图像等多模态数据，其中动作和观测状态均包含6个关节位置（肩部平移、肩部提升、肘部弯曲、腕部弯曲、腕部旋转和夹爪位置），前视图像为RGB视频（分辨率480x640，帧率30fps）。数据集总共有360个回合、150084帧和3个任务，数据以parquet格式存储，视频以mp4格式存储，总数据大小约300MB（数据100MB，视频200MB）。机器人类型为so_follower，适用于训练和评估机器人控制模型。This dataset was created using LeRobot for robotics applications. It includes multimodal data such as actions, observation states, and front-view images. Actions and observation states consist of 6 joint positions (shoulder pan, shoulder lift, elbow flex, wrist flex, wrist roll, gripper). Front-view images are RGB videos with a resolution of 480x640 and a frame rate of 30fps. The dataset contains 360 episodes, 150,084 frames, and 3 tasks. Data is stored in parquet format, with videos in mp4 format, totaling approximately 300MB (100MB for data, 200MB for videos). The robot type is so_follower, suitable for training and evaluating robot control models.