• Description:

Kuka iiwa peg insertion with force feedback

Split Examples
'train' 3,000
  • Feature structure:
    'episode_metadata': FeaturesDict({
    'steps': Dataset({
        'action': Tensor(shape=(4,), dtype=float32),
        'discount': Scalar(shape=(), dtype=float32),
        'is_first': bool,
        'is_last': bool,
        'is_terminal': bool,
        'language_embedding': Tensor(shape=(512,), dtype=float32),
        'language_instruction': Text(shape=(), dtype=string),
        'observation': FeaturesDict({
            'contact': Tensor(shape=(50,), dtype=float32),
            'depth_image': Tensor(shape=(128, 128, 1), dtype=float32),
            'ee_forces_continuous': Tensor(shape=(50, 6), dtype=float32),
            'ee_orientation': Tensor(shape=(4,), dtype=float32),
            'ee_orientation_vel': Tensor(shape=(3,), dtype=float32),
            'ee_position': Tensor(shape=(3,), dtype=float32),
            'ee_vel': Tensor(shape=(3,), dtype=float32),
            'ee_yaw': Tensor(shape=(4,), dtype=float32),
            'ee_yaw_delta': Tensor(shape=(4,), dtype=float32),
            'image': Image(shape=(128, 128, 3), dtype=uint8),
            'joint_pos': Tensor(shape=(7,), dtype=float32),
            'joint_vel': Tensor(shape=(7,), dtype=float32),
            'optical_flow': Tensor(shape=(128, 128, 2), dtype=float32),
            'state': Tensor(shape=(8,), dtype=float32),
        'reward': Scalar(shape=(), dtype=float32),
  • Feature documentation:
Feature Class Shape Dtype Description
episode_metadata FeaturesDict
steps Dataset
steps/action Tensor (4,) float32 Robot action, consists of [3x EEF position, 1x gripper open/close].
steps/discount Scalar float32 Discount if provided, default to 1.
steps/is_first Tensor bool
steps/is_last Tensor bool
steps/is_terminal Tensor bool
steps/language_embedding Tensor (512,) float32 Kona language embedding. See
steps/language_instruction Text string Language Instruction.
steps/observation FeaturesDict
steps/observation/contact Tensor (50,) float32 Robot contact information.
steps/observation/depth_image Tensor (128, 128, 1) float32 Main depth camera observation.
steps/observation/ee_forces_continuous Tensor (50, 6) float32 Robot end-effector forces.
steps/observation/ee_orientation Tensor (4,) float32 Robot end-effector orientation quaternion.
steps/observation/ee_orientation_vel Tensor (3,) float32 Robot end-effector orientation velocity.
steps/observation/ee_position Tensor (3,) float32 Robot end-effector position.
steps/observation/ee_vel Tensor (3,) float32 Robot end-effector velocity.
steps/observation/ee_yaw Tensor (4,) float32 Robot end-effector yaw.
steps/observation/ee_yaw_delta Tensor (4,) float32 Robot end-effector yaw delta.
steps/observation/image Image (128, 128, 3) uint8 Main camera RGB observation.
steps/observation/joint_pos Tensor (7,) float32 Robot joint positions.
steps/observation/joint_vel Tensor (7,) float32 Robot joint velocities.
steps/observation/optical_flow Tensor (128, 128, 2) float32 Optical flow.
steps/observation/state Tensor (8,) float32 Robot proprioceptive information, [7x joint pos, 1x gripper open/close].
steps/reward Scalar float32 Reward if provided, 1 on final step for demos.
  title={Making sense of vision and touch: Self-supervised learning of multimodal representations for contact-rich tasks},
  author={Lee, Michelle A and Zhu, Yuke and Srinivasan, Krishnan and Shah, Parth and Savarese, Silvio and Fei-Fei, Li and  Garg, Animesh and Bohg, Jeannette},
  booktitle={2019 IEEE International Conference on Robotics and Automation (ICRA)},