feat: add nyu franka play

[sdas-neuraco]

Features

• NYU Franka Play

Missing gripper values so the gripper visualization does not work.

[ypang-neuraco]

LGTM \w comments

[ypang-neuraco]

please update the description

[ypang-neuraco]

should these be visual joints?

[sdas-neuraco]

this dataset works out the box with the URDF so is it necessary?

[sdas-neuraco]

there is no gripper data at all so not sure how to proceed? as there is no way to hook up the grippers?

[ypang-neuraco]

I think unless we want to train on the finger joint values, its better to put it in visual joints.
You can leave visual_joint_type empty which will default to "CUSTOM" and you can log them like how you log joint positions.

On another note, perhaps we need functionality to go from joint position to gripper open amount in a separate PR.

from robot.py:

    def gripper_open_width_to_open_amount(
        self, gripper_open_width: float, gripper_joint_names: list[str]
    ) -> float:
        """Convert gripper joint positions to open amount.

        Args:
            gripper_joint_names: List of gripper joint names
            gripper_joint_positions: Dictionary mapping gripper
                joint names to position values

        Returns:
            Amount to open gripper (0-1)
        """
        joint_name = gripper_joint_names[0]  # Assume joints are symmetric
        min_pos, max_pos = self.joint_limits[joint_name]
        joint_pos = np.clip(gripper_open_width / 2.0, min_pos, max_pos)
        assert min_pos <= joint_pos <= max_pos, "Gripper joint position out of range"
        open_amount = (joint_pos - min_pos) / (max_pos - min_pos)
        assert 0.0 <= open_amount <= 1.0, "Gripper open amount out of range"
        return open_amount

[sdas-neuraco]

agreed we should make this a seperate PR would be good to have