videos are abailable :
https://www.youtube.com/watch?v=agjsPMJshy0
input1 : RGB-D colored PointCloud for scene input2 : non-colored .PCD (point cloud data) of target object
output : 6DoF estimation result in various ros style (tf flame, Pose, etc...).
in your catkin workspace,
git clone https://github.com/TsuruMasato/OnlineObjectDetector
cd online_object_detector
catkin build (or catkin_make)
- roslaunch openni2_launch openni2.launch
- rosparam set target_object_pcd_path your_object's_PCD_file
- rosrun online_object_detector online_object_detector
If you don't have PCD file, you can use PCD generator.
- rosrun online_object_detector pcd_generator your_3D_model output_file_name
and you can see the estimation result in Rviz, at camera flame.
I will make roslaunch file to make more convenient.
if you use this module for your research, business, or some other things, please cite our paper.
"Online Object Searching by a Humanoid Robot in an Unkown Environment" M.Tsuru et al. ICRA-RAL, 2021. https://ieeexplore.ieee.org/document/9361266