robokudo.io.ros_camera_without_depth_interface

ROS camera interface for RGB-only cameras in RoboKudo.

This module provides an interface for ROS cameras that only provide RGB images without depth information. It supports:

  • Synchronized RGB image and camera info subscription

  • Image rotation with intrinsics adjustment

  • Transform lookup between camera and world frames

  • Thread-safe data handling

The module is used for cameras like:

  • Hand cameras on robotic arms

  • Standard USB cameras with ROS drivers

  • Network cameras with ROS interfaces

Classes

ROSCameraWithoutDepthInterface

A ROS camera interface for RGB-only cameras.

Module Contents

class robokudo.io.ros_camera_without_depth_interface.ROSCameraWithoutDepthInterface(camera_config)

Bases: robokudo.io.camera_interface.ROSCameraInterface

A ROS camera interface for RGB-only cameras.

This class handles cameras that provide only RGB images without depth data. It synchronizes RGB images with camera calibration information and supports image rotation with proper intrinsics adjustment.

Variables:

  • color_subscriber – Subscriber for RGB image topic

  • cam_info_subscriber – Subscriber for camera info topic

  • color – Latest RGB image data

  • cam_info – Latest camera calibration info

  • cam_intrinsic – Camera intrinsic parameters in Open3D format

  • color2depth_ratio – Always (1,1) since no depth data

  • cam_translation – Camera translation from TF

  • cam_quaternion – Camera rotation as quaternion from TF

  • timestamp – Timestamp of latest data

  • lock – Thread synchronization lock

  • bridge – Bridge for converting between ROS and OpenCV images

color_subscriber
cam_info_subscriber
color = None
cam_info = None
cam_intrinsic = None
color2depth_ratio = None
cam_translation = None
cam_quaternion = None
timestamp = None
lock
bridge
rotate_camera_intrinsics(K: numpy.ndarray, image_size: tuple[int, int], rotation='90_ccw') → tuple[numpy.ndarray, tuple[int, int]]

Adjust camera intrinsics matrix for image rotation.

Computes the new camera intrinsics matrix and image dimensions after applying a rotation to the image. Supports 90° counter-clockwise, 90° clockwise, and 180° rotations.

Parameters:

  • K (numpy.ndarray) – Original camera intrinsic matrix of shape (3, 3)

  • image_size (tuple) – Original image dimensions as (width, height)

  • rotation (str, optional) – ‘90_ccw’, ‘90_cw’, ‘180’ (default is 90° counter-clockwise)

Returns:

Tuple of (new intrinsics matrix, new image dimensions)

Return type:

tuple(numpy.ndarray, tuple(int, int))

Raises:

ValueError – If rotation type is not supported

rotate_image_and_intrinsics(img, K, rotation='90_ccw')

Rotate an image and adjust its camera intrinsics accordingly.

Parameters:

  • img (numpy.ndarray) – Input image to rotate

  • K (numpy.ndarray) – Camera intrinsics matrix

  • rotation (str, optional) – Type of rotation to apply, defaults to ‘90_ccw’

Returns:

Tuple of (rotated image, new intrinsics matrix, new dimensions)

Return type:

tuple(numpy.ndarray, numpy.ndarray, tuple)

Raises:

ValueError – If rotation type is not supported

callback(color_data: sensor_msgs.msg.Image, cam_info: sensor_msgs.msg.CameraInfo | None = None)

Process synchronized RGB image and camera info messages.

TODO make this generic. handle the encoding and order properly. For standard and compressed images. this might also depend on the fix of image_transport_plugins being published as a package. Startpoint can be found at the bottom of this method.

Parameters:

  • color_data (sensor_msgs.msg.Image) – RGB image message

  • cam_info (sensor_msgs.msg.CameraInfo, optional) – Camera calibration info message, defaults to None

set_data(cas: robokudo.cas.CAS)

Update the Common Analysis Structure with latest camera data.

This method: * Applies any configured image rotation * Updates camera intrinsics accordingly * Sets RGB image, camera info, and transform data in the CAS * Handles thread synchronization

Parameters:

cas (robokudo.cas.CAS) – Common Analysis Structure to update