Stereolabs ZED Camera - Alvar AR tag example

This package shows how to use the ZED Wrapper with ar_track_alvar to detect AR markers and localize them in the space.

The ar_track_alvar package is a ROS wrapper for Alvar, an open source AR tag tracking library.

Installation

Follow the install guide

Run the program

To launch the example, open a terminal and launch:

$ roslaunch zed_ar_track_alvar_example zed_ar_track_alvar.launch

The launch file automatically starts the zed_wrapper node, the ar_track_alvar node and RVIZ with a preconfigured view:

• the Left Image view shows the RGB stream from the left camera of the ZED.
• the Left Camera view shows the “world” as seen by the left camera. The TF and the Tags are selected to be shown in 3D projection
• the 3D view shows the ZED camera localized in the space and the position of the Tags as coloured markers.

Note It is important to subscribe the /zed/pose topic to be able to localize the camera in the space.

Note the tags used in the example are available following this link

The launch file explained

To use the ZED wrapper with the ar_track_alvar node we need to correctly set the following ar_track_alvar parameters:

• camera_image -> the name of the topic that provides camera frames for detecting the AR tags.
• camera_info -> the name of the topic of camera parameters used to correctly localize the tag in the 3D world
• output_frame -> the name of the TF frame associated to the camera position
• camera_model -> model of the ZED camera: zed for ZED, zedm for ZED Mini, zed2 for ZED 2

It is important that the zed_wrapper node and the ar_track_alvar node use the same values for these three parameters to be able to correctly communicate.

The values associated to the above parameters are the following:

    <arg name="cam_image_topic"         default="rgb/image_rect_color" />
    <arg name="cam_info_topic"       	default="rgb/camera_info" />
    <arg name="camera_frame"            default="zed_left_camera_frame" />

The corresponding parameters of the ZED node are the following:

• cam_image_topic -> rgb_topic
• cam_info_topic -> rgb_info_topic
• camera_frame -> left_camera_frame

The corresponding parameters of the rtabmap_ros node are the following:

• cam_image_topic -> camera_image
• cam_info_topic -> camera_info
• camera_frame -> output_frame

It is also important to set the correct marker_size in centimeters. Better if the real size of the markers is measured after printing them:

<arg name="marker_size" default="5.0" />

Other parameters that can be tuned:

• max_new_marker_error -> A threshold determining when new markers can be detected under uncertainty
• max_track_error -> A threshold determining how much tracking error can be observed before an tag is considered to have disappeared

Their values in the launch file:

<arg name="max_new_marker_error" default="0.08" />   
<arg name="max_track_error" default="0.2" />

Markers

Following two examples of markers with ID numbers from 0 to 17: