Robotic Hand-eye Calibration with Depth Camera Using Sphere-Model

This repository is the backup for my part @ ** JingDong GRASPING ROBOT CHALLENGE ** Also the implements of my paper: ** Robotic hand-eye calibration with depth camera: A sphere model approach **

| Ubuntu 1604 & ROS Kinetic Kame |

Related Publications:

Yang, Lixin, et al. " Robotic hand-eye calibration with depth camera: A sphere model approach. " 2018 4th International Conference on Control, Automation and Robotics (ICCAR). IEEE, 2018. PDF

@inproceedings{yang2018robotic,
  title={Robotic hand-eye calibration with depth camera: A sphere model approach},
  author={Yang, Lixin and Cao, Qixin and Lin, Minjie and Zhang, Haoruo and Ma, Zhuoming},
  booktitle={2018 4th International Conference on Control, Automation and Robotics (ICCAR)},
  pages={104--110},
  year={2018},
  organization={IEEE}
}

License

handeyeCalibWithDepthCamera is freely available for free non-commercial use, and may be redistributed under these conditions. Please, see the license for further details

Prerequisit

Ubuntu 16.04
ROS Kinetic Kame // full install
**PCL 1.7+ ** (should be included in ROS Kinetic Kame)
OpenCV 3.3 (should be included in ROS Kinetic Kame)
cv_bridge
sudo apt-get install ros-kinetic-cv-bridge
Eigen3 (should be included in Ubutnu 16.04)
sudo apt-get install libeigen3-dev
Sophus (https://github.com/stonier/sophus) // standard make && make install
Ceres (http://ceres-solver.org/) // follow the installation guild
librealsense2 (https://github.com/IntelRealSense/librealsense) :
We use the librealsense2 version to drive Intel RealSense ZR300 Camera. and wrap it by ROS topics and services. Our realsense2_driver need the librealsense2.so. The recommended way to install it is by following the Linux guide:(https://github.com/IntelRealSense/librealsense/blob/master/doc/distribution_linux.md) , or as below:
Register the server's public key :
sudo apt-key adv --keyserver keys.gnupg.net --recv-key C8B3A55A6F3EFCDE || sudo apt-key adv --keyserver hkp://keyserver.ubuntu.com:80 --recv-key C8B3A55A6F3EFCDE
Add the server to the list of repositories in Ubuntu 16 LTS:
sudo add-apt-repository "deb http://realsense-hw-public.s3.amazonaws.com/Debian/apt-repo xenial main" -u
Then:
sudo apt-get update
And install by apt-get :
sudo apt-get install librealsense2-dev

Compile `calib_ws`

standard ROS workspace catkin_make procedure;

cd ~/WhereEver_You_Want
git clone https://github.com/lixiny/calib_ws.git 
cd calib_ws	
catkin_make

Usage

1. Plug-In the Realsense Camera & Open a Terminal 1:

cd calib_ws
source devel/setup.bash
roslaunch realsense2_driver realsense_driver.launch

when it shows the INFO like:

means the realsense2_driver is successfully tuned up.

Important: you should use the pop-up parameters: ###### camera_intrinsics ###### to update: ./calib_ws/src/handeye_calib_sphere/camera_intrinsic_color.xml .

After roslaunch, you should have topics as blow in your terminal when you run: rostopic list

2. Prepare the Calibration Set-ups :

2.1 A Shpere Model:

we use a 3D-printed hollow sphere model with Radius 75mm ( you can modify this radius if necessary).

2.2 An Aruco Plate (not compulsory ):

we use RANSAC to localize the sphere inside the captured pointclouds. instead to iterate from a full bunch of scatted points, if we can use a aruco plate to segmente a region and make sure the points inside the region are all captured on the sphere, RANSAC would be much easier( it should be ).

The Aruco Plate is not compulsory again, you can de-activate the aruco module as described in Advanced Usage part.

We use the aruco plate with 2 different size:

2.3 Eye-in-hand Camera Mounting:

Till now, we only implement the eye-in-hand configration. If you are familiar with the hand-eye problem, you can easily implement the eye-to-hand yourself.
We use the hand-mounted camera viewing the Sphere-ArucoPlate setups as shown below. [这里添一张图片]

2.4 Gave a Initial Value

3. Tune up Robot, record the TFlink name of `base_link` and `end_link`

we use UR5 and self-defined [我还没写完]

5. Open another termianl 2:

cd calib_ws
source devel/setup.bash
## 开始采集数据:

TienPhuocNguyen/calib_ws_cpp