zhangbaozhe / icg_ros

A ROS wrapper for Iterative Corresponding Geometry (ICG)

Hi there 👋

I am Baozhe, currently an undergraduate student at The Chinese University of Hong Kong, Shenzhen. My major is computer engineering. I am interested in developing highly efficient real-time robot algorithms (using C++) that can run with limited resources (e.g., edge computers on robots). I am interested in multi-robot systems, motion planning, SLAM, optimal control, and machine learning with robots.

🤔 Recent news:

• 🔭 The source code of CoNi-MPC is released! Please check CoNi-MPC for more information.
• 🌱 I finished my RA internship at Fast Lab, ZJU (2022/06 - 2023/08).
• ⚡ I am learning reinforcement learning.
• ⚡ I am learning computer graphics.

📫 How to reach me:

• My school email: [email protected]
• My personal email: [email protected]

icg_ros's Issues

More information about the proper setup is needed

Hi!

I was trying out your wrapper package, but I am not able to understand what is the expected behavior. I changed the color_camera_topic, color_camera_info_topic, depth_camera_topic, and depth_camera_info_topic and see that the node subscribes to them.
Additionally, Viewer already exists message and GUI with the image topic appears.

However, I am not sure how to proceed further. Can you provide more information? I see that the node expects to get bool messages on the /start_tracking topic, and I publish to it and notice no effect. I also subscribe to the /pose topic which is not publishing anything. Finally, the temp_transform does not seem to change at all (I show a triangular shape to camera).

Segmentation fault (core dumped)

Hi, i'm having an issue, after running rosrun icg_ros icg_test_node -config_dir=src/icg_ros/config i'm getting the next output on terminal:

Viewer already exists Segmentation fault (core dumped)

How can I fix this? there is no GUI showing, just that

Detailed steps to run the code

Hello, thanks a lot for your repo. It it a little complicated for me to run the orignal repo by DLR. But I am familiar with ROS and could be able to run your code well.

I would like to ask some questions:
(1) I am not sure if I am running the repo correctly. My goal is to track some objects from YCB dataset. So, I downloaded the .obj files here and put textured.obj under the config/. Copy and paste the xx.yaml and xx_detector.yaml. Then, put the according object in front of the camera and run rosrun icg_ros icg_test_node -config_dir=icg_ros/config -camera_frame=camera_link, which would creat two *_model.bin files and launch the gui. Finally, press D to initilize and T for tracking.
(2) For the pose published in ROS, if I keep a identity transformation matrix in triangle.yaml, it means that it is the original/zero frame where the object is shown in MeshLab, with respect to the camera_frame inputed to icg_test_node?
(3) The transformation matrix in triangle_detector.yaml seems to be the initial estimation pose for alignment? If there is a big initial error, it would be hard to track the object. So, it it better to make sure that the initial pose is almost correct?
(4) How could I do if I would like to track multiple objects? As long as we have the .obj files then we could be able to track them?

Thanks you again!

Is it possible to run this package only with RGB images?

Hi,
I'm wondering if this can be run without depth information or not?