Ask Your Question

ar_track_alvar not precise and not producing repeatable results

asked 2018-09-27 08:32:11 -0600

Sufyan gravatar image

updated 2018-10-04 11:39:38 -0600

In my project i am taking a picture from an IP camera and converting that image in to a rostopic.. Then i used ar track to get information from visualization_marker topic and its working very well but the problem is that sometimes the Tag is not recognized by ar track and thus it publishes nothing on visualization marker ... all I want from ar track is to give me orientation of my tag..but out of 5 images only 3 will produces correct result sometimes even 2 .. following is the launch file for ar track can anyone help me with parameters i have put the marker size in cm... is it the correct unit? and the other two parameters max_new_marker_error and max_track_error should i set them to obtain optimum results. and also please tell me if ar_track_alvar consistently produces un authentic results ...What could be the other factors that can effect the precesion of Ar_track_alvar

    <arg name="marker_size" default="12"        />
    <arg name="max_new_marker_error" default="0.08" />
    <arg name="max_track_error" default="0.2" />
    <arg name="cam_image_topic" default="/Sufyan_ImageTopic" />
    <arg name="cam_info_topic" default="/v4l/camera/camera_info" />
    <arg name="output_frame" default="/v4l_frame" />
    <node name="ar_track_alvar" pkg="ar_track_alvar" type="individualMarkersNoKinect" respawn="false" output="screen">
        <param name="marker_size"           type="double" value="$(arg marker_size)" />
        <param name="max_new_marker_error"  type="double" value="$(arg max_new_marker_error)" />
        <param name="max_track_error"       type="double" value="$(arg max_track_error)" />
        <param name="output_frame"          type="string" value="$(arg output_frame)" />
        <remap from="camera_image"  to="$(arg cam_image_topic)" />
        <remap from="camera_info"   to="$(arg cam_info_topic)" />



edit retag flag offensive close merge delete



Your launch file looks okay, can you upload the camera images somewhere so we can see them. It's possible the image quality is too low for the tags to be reliably detected.

PeteBlackerThe3rd gravatar image PeteBlackerThe3rd  ( 2018-09-27 10:44:39 -0600 )edit

Hey!! sorry for late reply. I am using a 1 MP camera and my ar tag is about 4 meters from camera .. I have edited the question so now it contains 2 images

Sufyan gravatar image Sufyan  ( 2018-10-04 11:37:54 -0600 )edit

1 Answer

Sort by ยป oldest newest most voted

answered 2018-10-04 11:52:02 -0600

I have an idea why the tag detection is not working as well as it could be.

The white border areas around the tag is very important for it to be detected robustly, ideally you want this to be at least 2 squares of the pattern wide. Also if you could make sure the tag and the border are kept as flat as possible this will help to remove inconsistent shadowing artefacts.

Your second image has pretty much no white border at all, so I wouldn't expect that to be detected very well at all.

Hopefully this helps you get this working reliably.

edit flag offensive delete link more


can the camera callibration be the problem ? because besides all the working i get a small error on this node which is as below

[ERROR] [1538676611.129294992]: "v4l_frame" passed to lookupTransform argument target_frame does not exist.

how can I solve this probelm?

Sufyan gravatar image Sufyan  ( 2018-10-04 13:38:25 -0600 )edit

Your Answer

Please start posting anonymously - your entry will be published after you log in or create a new account.

Add Answer

Question Tools

1 follower


Asked: 2018-09-27 08:32:11 -0600

Seen: 418 times

Last updated: Oct 04 '18