@Boyuan-Tian I meet the same problem after I clone the code and `make`, `./pc_pipeline` with the default config.txt

[hanm2019]

@Boyuan-Tian I meet the same problem after I clone the code and make, ./pc_pipeline with the default config.txt

the version is :

os: ubuntu 18.04
pcl: 1.12
gcc: 7.5.0

Originally posted by @hanm2019 in #2 (comment)

[Boyuan-Tian]

Hi @hanm2019,

Thanks for your interest!

I saw you are using the provided sample data but PCL 1.12. My guess is that this might be a version issue since we didn’t encounter this on PCL 1.7/1.8.

I will try to reproduce the issue by running on 1.12 later this week since I have been in a very busy final week recently.

I will get back to you when I have any updates.

Best regards,
Boyuan Tian

[hanm2019]

Thanks

[Boyuan-Tian]

Hi @hanm2019 ,

I tested on PCL 1.12 and reproduced this segfault error. It is caused by the upgraded PCL library.

This repo contains our approximation mechanism that is implemented by hacking the standard PCL library, so we uploaded the modified files (.h and .hpp) to “./pcl-1.7/pcl”. Our executable is linked to the shared objects (.so) compiled from PCL (.cpp), so if there are essential discrepancies between *.hpp and *.cpp files, PCL’s implementation may not be correctly executed. In this case, *.h and *.hpp files are in PCL 1.7, while *.so files are compiled from PCL 1.12.

We primarily tested the code on PCL 1.7/1.8/1.9, so we recommend using the above versions.

If you prefer to stick on 1.12, you can switch to HARRIS and SIFT keypoint detection algorithm by changing the corresponding line (Ln #12) in config.txt, as this issue should only affect NARF key points so far due to critical updates.

We plan to fix this issue when time permits, so do a vanilla pipeline independent of particular library versions.

Please let me know if you have further questions.

Best regards,
Boyuan Tian

[hanm2019]

Thank you for your response. I will try PCL1.9 instead of 1.12.

[HansonHou798]

I encountered the same problem and solved the bug after reading your answer. Thank you for your open source work!