warned_keys is a trick (hack) to have a global variable to prevent repeated warnings. will work as long as it remains the default val

Hollow cube of side 2*half_extent centered at the origin, six faces sampled at `spacing`.

Single-instant timestamps for the second scan: deskewing reduces to identity when every point shares end_time.

Logs a vector as three scalar time-series in rerun. Args: rerun: rerun module entity_path_prefix: Base path for scalar logs

Add IMU measurement to pipeline (will be buffered until processed by lidar). Parameters ---------- time : int

Same as feed_imu but with per-sample Gaussian noise added to accel and gyro. Seed is fixed at the call site so the test is deterministic.

constants and util funcs

Read the frame_id from the first message of the given topic.

return the 4x4 transform from the base frame to the world/odometry frame

Convert [qx, qy, qz, qw, x, y, z] to 4x4 transform.

Compute transform matrix from 'from_frame' to 'to_frame' using a static TF tree.

Reset all accumulated statistics to zero. */

Convert 4x4 transform matrix to [qx, qy, qz, qw, x, y, z].

if true, there's at least one valid lidar and imu data sequence

Load IMU->base and LIDAR->base transforms for HeLiPR.

reads the next data point and appends to the appropriate buffer

Because for some reason, I cannot use a raw bool in configurable parameters dict because of how DeclareLaunchArgument seems to work, so i hav

Helper for aeva: get file stem as int64_t

Build a static TF tree from only '/tf_static' messages. Returns: dict: { child_frame_id: (parent_frame_id, transform) }

Convert a Iterable of sensor_msgs.msg.PointField messages to a np.dtype. :param fields: The point cloud fields. (Type: ite

Write LIO results to disk under LIOPipeline.output_dir. Writes: - Trajectory (timestamps and poses) in TUM format text file.

Return only CLI parameters that were explicitly set by the user