apriltag2_detection

This package is used to analyze the relative pose of single duckiebot estimated by using AprilTag2. Except for the normal pipeline of AprilTag2, it also provides post-processing srcipts to get result analysis, as well as several unit-tests for testing the correctness of the system.

result analysis

This package provide detection_post_process.launch to analyze the performance of relative pose estimation by using AprilTag2. Each time it collects a certain number of estimated pose (both postion and rotation) and then compute the following:

• max, min, range, mean, variance of these pose estimation.
• time used for each subprocess in the pipeline of relative pose estimation (detecting AprilTag + computing relative pose).
• cpu/ram comsumption for relative pose estimation.

=======

1. catkin_make
2. source devel/setup.bash
3. roslaunch pi_camera camera_apriltag_demo.launch veh:=duckiebot
4. roslaunch apriltags2_ros apriltag2_demo.launch veh:=duckiebot
5. roslaunch apriltags2_ros detection_to_local_frame.launch veh:=duckiebot
6. rosparam set enable_statistics true & rosrun rosprofiler rosprofiler
7. roslaunch apriltags2_ros detection_post_process.launch veh:=duckiebot

unit test

Several test are provided in this package for testing the correctness of the system.

detection_to_local_frame_testor_node.test

1. This is used to test the correctness of pose estimation. All images in tests/test_image are tested and the name of the images are their groundtruth in degree. If you want to specify one image, for now you should put only that image in the above folder.
2. In the test, each time we publish one of these images as well as camera info. apriltag_detector_node is launched to detect apriltag and publish pose estimation (camera frame with respect to tag frame). detection_to_local_frame is launched for coverting pose estimation into appropriate frame (robot wrt world), which is the ultimate estimation of pose we are using for tests.

cd [REPO_ROOT]
catkin_make

source act_addon_environment.sh

rostest apriltags2_ros detection_to_local_frame_testor_node.test

• rostest --text apriltags2_ros detection_to_local_frame_testor_node.test to get console output to the screen

Duckietowm-Spesific

April Tag detection pipeline uses two container: the base container and the apriltag-addon container. This documemt prepend each command with the name of the environment in which that command needs to be run, i.e. ´[LOCAL_PC]´ represents that the command should be run on your local PC. Make sure to remove the environment identifier ´[ENVIRONMENT_NAME]´ before executing the command.

First set the DOCKER_HOST on your terminal

[LOCAL_PC] export DOCKER_HOST=![ROBOT_NAME].local

Then, launch the base container by executing

[LOCAL_PC] docker -H ![ROBOT_NAME].local run -it --net host --privileged -v /data:/data --memory="300m" --memory-swap="2.0g" --name base selcukercan/frpi-base:v5 /bin/bash

While in the base container roslaunch the file that starts the camera, decoding and rectification nodes

[BASE_CONTAINER] roslaunch pi_camera camera_apriltag_demo.launch veh:=![ROBOT_NAME]

Then open up a new container and set ´DOCKER_HOST´ in this window as well.

[LOCAL_PC] export DOCKER_HOST=![ROBOT_NAME].local

In the new terminal window, launch the apriltag-addon container

[LOCAL_PC] docker -H ![ROBOT_NAME].local run -it --net host --privileged -v /data:/data --memory="300m" --memory-swap="2.0g" --name base selcukercan/apriltag2add-on:v12 /bin/bash

Now you will be inside selcukercan/apriltag2add-on. Next, start the apriltag core algorithm

[APRILTAG_ADDON_CONTAINER] roslaunch apriltags2_ros apriltag2_demo.launch veh:=mete

At this point you should be able to see the results produced by the apriltag algorithm by inspecting

[APRILTAG_ADDON_CONTAINER] rostopic echo /![ROBOT_NAME]/tag_detections

Note that this returns the pose of the camera frame with respect to the apriltag frame. To learn more about the coordinate frame definitions/conventions, please refer to this.

To get the position in fixed frame XYZ Euler Angles

[APRILTAG_ADDON_CONTAINER]  roslaunch apriltags2_ros detection_to_local_frame.launch veh:=![ROBOT_NAME]

To inspect the orientation expressed in Euler Angles

[APRILTAG_ADDON_CONTAINER] rostopic echo ![ROBOT_NAME]/apriltags2_ros/publish_detections_in_local_frame/tag_detections_local_frame

Documentation

This project uses sphinx to generate project documentation automatically.

To start generating documents you have to first install sphinx

sudo apt-get install python-sphinx

generate the documentation

cd ![APRILTAG_ADDON_ROOT]/src/apriltags2_ros/apriltags2_ros/docs
make html

click to open the generated document which is located at ![APRILTAG_ADDON_ROOT]/fapriltag2_addon/src/apriltags2_ros/apriltags2_ros/docs/build/html