Must return a dictionary with the following structure: { "Hips": (position, orientation), "Spine": (position, orientation),

Initialize the Xsens device and discover available body parts. Returns: True if initialization successful, False

Load robot motion data from a pickle file.

Normalizes a tensor over some axis (axes) :param x: data tensor :param axis: axis(axes) along which to compute the norm :param eps:

the pos offsets are applied in the local frame

解析 JSON 数据到窗口的 offsets 字典格式 {(joint_idx, channel_idx): offset}。

Performs quaternion multiplication on arrays of quaternions :param x: tensor of quaternions of shape (..., 4) :param y: tensor of quatern

Reads a BVH file and extracts animation information. :param filename: BVh filename :param start: start frame :param end: end frame

Initialize legacy support for current frame index

Unpacks frame suffix data if the major case is 0

Unpacks frame suffix data from inclusive of NatNet 2.7 to but not including NatNet 3

Unpacks frame suffix data inclusive from NatNet 3 to NatNet 4

Unpacks frame suffix data from NatNet 4.1 to present NatNet

Unpacks frame suffix data for any NatNet version before NatNet 2.7

Unpack marker description packet

Calculates offset for case where major version is 0

Calculates offset starting at NatNet 2.6 and going to (but not inclusive of 3)

Calculates offset for NatNet 3 and above for rigid body unpacking

Helper function for NatNet 0 case

Helper function for NatNet version inclusive 2, to not inclusive 3

Helper function for NatNets versions 3 to 4.0 not inclusive of 4.0

Unpack rigid body helper function for NatNet 4.2

Unpack rigid body description data for NatNet Versions 4 and 4.1

Helper function for NatNet versions under 2

Calculates offset for anything below NatNet 2.6

Normalize the initial pelvis yaw so all recordings start facing forward. On the first frame, captures the pelvis yaw and computes it

Build dictionary of available body mappings.

Check which required bodies are missing. Returns: List of missing required body names

Ensure quaternion is in scalar-first format [w, x, y, z]. Args: quat: Quaternion array (either [w,x,y,z] or [x,y,z,w])

Find the nearest mocap frame for a given video frame timestamp

Specifies which joints, actuators, and equality constraints should be removed from the Mujoco specification. Returns: Tu

Load MP4 and TXT data, then preprocess all frames

Load MP4 video file

Load and parse tracking data from TXT file

Preprocess all tracking data for all frames with timestamp alignment

Process body data for a single frame

Process controller data for a single frame

Process headset pose for a single frame

Process left hand data for a single frame with fallback to previous frame

Process right hand data for a single frame with fallback to previous frame