spco2_boo

This repository is SpCoSLAM (added Bag-of-Objects) package.
SpCoSLAM learns the relationship between object and place.
Mainly, it is a program that executes SpCoSLAM in Gazebo.

• Maintainer: Shoichi Hasegawa ([email protected]).
• Author: Shoichi Hasegawa ([email protected]).

You can do it like this below image.
(You need to prepare room layouts personally.)

Content

• Execution Environment
• Execute Procedure
• Folder
• To Do
• Reference
• Acknowledgements

Execution environment

• Ubuntu：20.04LTS
• ROS：Noetic
• Python：3.8.10
• C++：14
• Robot：Human Support Robot (HSR)
• YOLOv5 (Emergent Systems Lab original)

Execution Procedure

1 cd HSR/catkin_ws/src
2. git clone https://github.com/Shoichi-Hasegawa0628/spco2_boo.git
3. cd ~/HSR/ && bash ./RUN-DOCKER-CONTAINER.bash
4. cd catkin_ws
5. catkin_make
6. Launch Gazebo
7. Launch Rviz
8. Launch YOLOv5
9. rosnode kill /pose_integrator
10. roscd rgiro_spco2_slam
11. cd bash
12. bash reset-spco2-slam-output.bash
13. roslaunch rgiro_spco2_slam spco2_slam.launch
14. roslaunch rgiro_spco2_slam spco2_word.launch

Teaching the place name while teleoping with rqt.

Folder

• rgiro_openslam_gmapping：SpCoSLAM Wrapper of FastSLAM2.0 published on OpenSLAM
• rgiro_slam_gmapping：SpCoSLAM Wrapper of slam_gmapping ros package (ros wrapper of openslam_gmapping)
• rgiro_spco2_slam：Main codes of SpCoSLAM
• rgiro_spco2_visualization：Visualization codes of learning spatial concepts

To Do (Japanese)

• 絶対パスが含まれている箇所があるため、変更が必要な場合あり

Reference

• SpCoSLAM 2.0