taccap-g1-Coil-secure-cable-with-Velcro-0724

该数据集是使用LeRobot框架创建的机器人数据集，专注于双TacCap夹爪机器人（bi_taccap_gripper）的操作任务。数据集包含214个episodes，总计373,294帧，数据以parquet文件和视频格式存储，帧率为30fps。特征包括20维的动作和观察状态向量（涵盖左右夹爪的TCP位置、旋转和夹爪位置），以及多个视频观察：左右触觉传感器图像（分辨率400x700）和左右手腕摄像头图像（分辨率480x640）。数据集用于机器人学习任务，如强化学习或模仿学习，以支持触觉反馈的机器人控制。This robotic dataset was developed using the LeRobot framework, focusing on manipulation tasks for the bi_taccap_gripper dual TacCap gripper. It consists of 214 episodes, totaling 373,294 frames, and is stored in both Parquet file format and video format with a frame rate of 30 fps. The dataset includes 20-dimensional action and observation state vectors, which cover the TCP positions, rotations, and gripper positions of both the left and right grippers, alongside multiple video observations: left and right tactile sensor images with a resolution of 400×700, and left and right wrist camera images with a resolution of 480×640. This dataset is designed for robotic learning tasks such as reinforcement learning or imitation learning, to support robotic control with tactile feedback.