6D Pallet Pose Estimation (D6.4 ILIAD)

This contains a brief guide how to install / run the ROS-based simplified Object-RPE developed at ORU in D6.4 ILIAD project for pallet detection and pose estimation.

The EUR/EPAL-pallet is 1,200 mm × 800 mm × 144 mm

Installation and compile the source

The tools require full ROS installation. The installation assumes you have Ubuntu 16.04 LTS [ROS Kinetic]

1. Clone the repository and switch to iliad branch

   $ https://github.com/hoangcuongbk80/Object-RPE.git
   $ git checkout iliad

2. ROS

   $ cd ~/catkin_ws
   $ catkin_make install

3. Segmentation here
4. 3D mapping here
5. 6D object pose estimation here

Testing

1. Download data folder here and copy to ~/catkin_ws/src/Object-RPE

2. To test segmenation module:

   $ cd ~/catkin_ws/src/Object_RPE/Mask-RCNN/samples/warehouse
   $ python3 eval.py

   It will read rgb and ground-truth images from /rgb and gt/ folders in .../Object-RPE/data then save results (mask and accuracy.txt file) into .../Object-RPE/data/mask

   

3. To test 3D mapping module:

   $ cd ~/catkin_ws/src/Object-RPE/obj_pose_est/mapping/app
   $ ./obj_pose_est/mapping/app/build/mapping -l data/ 100

   

4. To test pose estimation module (Video):

   $ roscore

   In an other terminal:

   $ rosrun obj_pose_est ObjectRPE_srv.py

   In an other terminal:

   $ roslaunch obj_pose_est launch_object_rpe.launch

   

How to operate the system?

$ roscore
$ rosrun obj_pose_est ObjectRPE_srv.py
$ roslaunch obj_pose_est launch_rpe_cam.launch
$ roslaunch openni2_launch openni2.launch