ROS Kinetics + Realsens D435i + ORK + LINEMOD Object Recognition

1. ORK

Website: https://wg-perception.github.io/object_recognition_core/

ORK (Object Recognition Kitchen) is an ROS integrated object recognition library. Current Kinetic version of ROS only integrates some binary installation files of feature packages, so it needs to be compiled and installed through source code.

Install Dependent Libraries

sudo apt-get install meshlab
sudo apt-get install libosmesa6-dev
sudo apt-get install python-pyside.qtcore
sudo apt-get install python-pyside.qtgui

Create workspace, download feature pack source, compile

mkdir ork_ws && cd ork_ws
wstool init src https://raw.github.com/wg-perception/object_recognition_core/master/doc/source/ork.rosinstall.kinetic.plus
cd src && wstool update -j8
cd .. && rosdep install --from-paths src -i -y
catkin_make

Setting environment variables

echo "export ~/ork_ws/devel/setup.bash" >> ~/.bashrc
source ~/.bashrc

2. CouchDB Building Model Library

The LINEMOD algorithm in ORK is based on template matching. It needs to build a data model of known objects, match one by one according to the collected information, and find the matching objects in the database.First, create the database with the CouchDB tool:

install

sudo apt-get install couchdb

Test whether the installation was successful, as shown in the diagram, indicating that the installation was successful

curl -X GET http://localhost:5984

Data for creating a Coca-Cola can model in a database

rosrun object_recognition_core object_add.py -n "coke " -d "A universal can of coke" --commit

View in Browser http://localhost:5984/_utils/database.html?object_recognition/_design/objects/_view/by_object_name , copy the id (a string of numbers like scrambles), and use
Loading a 3D model of a Coke can, including a coke.stl model in ork_tutorials, downloads it directly.Download ork_tutorials

git clone https://github.com/wg-perception/ork_tutorials

Loading the coke.stl model into the data

rosrun object_recognition_core mesh_add.py bb01ae7a23033bdef1a1c3b76000092c ~/ork_ws/src/ork_tutorials/data/coke.stl --commit

Open the above web address in the browser again, as shown in the following image

Install the couchapp tool to view specific models in the browser

sudo pip install git+https://github.com/couchapp/couchapp.git
rosrun object_recognition_core push.sh

View in browser

3. Model Training

rosrun object_recognition_core training -c `rospack find object_recognition_linemod`/conf/training.ork

The training is completed as shown in the following figure

4. Object recognition

Start realsense d435i camera

roslaunch realsense2_camera rs_camera.launch filters:=pointcloud

Reference resources: https://github.com/IntelRealSense/realsense-ros

Run the following command for object detection

rosrun object_recognition_core detection -c  `rospack find object_recognition_linemod`/conf/detection.ros.ork

We found that detection.ros.ork subscribed to the following topics:

The corresponding topic published by realsense is:

/camera/depth/camera_info
/camera/depth/image_rect_raw
/camera/color/camera_info
/camera/color/image_raw

So you need to use topic_tools relay for topic mapping, as follows:

rosrun topic_tools relay /camera/depth/camera_info    /camera/depth_registered/camera_info
rosrun topic_tools relay /camera/depth/image_rect_raw    /camera/depth_registered/image_raw
rosrun topic_tools relay /camera/color/camera_info    /camera/rgb/icamera_info
rosrun topic_tools relay /camera/color/image_raw   /camera/rgb/image_rect_color

But mapping each time is too cumbersome, just change the topic name of the detection.ros.ork subscription:

gedit `rospack find object_recognition_linemod`/conf/detection.ros.ork

Modify as follows:

After you have mapped the topic or set up the above, run the detection command again:

rosrun object_recognition_core detection -c  `rospack find object_recognition_linemod`/conf/detection.ros.ork

The following information appears:

Open rviz view
- Add Pointcloud 2, select topic/camera/depth/color/points
- Add OrkObject, select topic/recognized_object_array (this topic will be published if it is successfully identified)