couldn't read PipelineParams.yaml file with segmentation fault

[PengZai]

Description:
I have successfully built the whole ros project without any error come out.

but when I run roscore, and then run

rviz -d $(rospack find kimera_vio_ros)/rviz/kimera_vio_euroc.rviz
roslaunch kimera_vio_ros kimera_vio_ros_euroc.launch

it will occurs a segmentation fault with reading .yaml.

I suspect it is a problem with opencv, so I write a small exmaple mimic with the implement of YamlParser.h, it works, I could read the .yaml file with CHECK_NOTNULL(fs)->open(filename, cv::FileStorage::READ);

however, In Kimera-VIO-ROS, we just couldn't read this catkin_ws/src/Kimera-VIO/params/Euroc/PipelineParams.yaml
there is only one Opencv in my PC, that is 3.4.0, I have check the version of opencv when I build them, both are 3.4.0

Command:

roslaunch kimera_vio_ros kimera_vio_ros_euroc.launch

process[kimera_vio_ros/kimera_vio_ros_node-1]: started with pid [1078237]
process[kimera_vio_ros/posegraph_viewer-2]: started with pid [1078238]
[ INFO] [1733959322.787804313]: Initializing pose graph visualizer
================================================================================REQUIRED process [kimera_vio_ros/kimera_vio_ros_node-1] has died!

Additional files:
Please attach all the files needed to reproduce the error.

Please give also the following information:

• KimeraVIO branch, tag or commit used
• GTSAM version used: 4.2.0
• OpenGV version used:
• OpenCV version used: type opencv_version 3.4.0
• Operating system and version (e.g. Ubuntu 16.04 or Windows 10): Ubuntu 20.04
• Did you change the source code? (yes / no): no

[PengZai]

this is my small example, it works in environment of Opencv 3.4.0

#include <opencv2/core.hpp>
#include <iostream>
#include <string>
#include <glog/logging.h>


using namespace std;
 
 
int main(int argc, char* argv[])
{
    

    google::InitGoogleLogging(argv[0]); 

    FLAGS_log_dir = "./logs";

    std::string filename = "/home/zhipeng/catkin_ws/src/Kimera-VIO/params/Euroc/PipelineParams.yaml";
    cv::FileStorage fs_;

    cv::FileStorage *fs = &fs_;
    CHECK_NOTNULL(fs)->open(filename, cv::FileStorage::READ);
    // fs->open(filename, cv::FileStorage::READ);
    std::cout << fs->isOpened() << std::endl;
    if (!fs->isOpened())
    {
        std::cout<< "OK" << std::endl;
    }
    LOG(INFO) << "Found " << 11 << " cookies"; 
    VLOG(1) << "I’m printed when you run the program with --v=1 or higher";
    LOG(INFO) << "Cannot open file: " << '\n';
    
    std::string id = "backend_type";

    float a = fs_["backend_type"];

    std::cout << a << std::endl;

    
}

[PengZai]

small update, I even copy the reading section in Kimera-VIO-ROS to my code, it won't occur segmentation fault.

#include <opencv2/core.hpp>
#include
#include
#include <glog/logging.h>
#include "yamlPasrser.h"

using namespace std;

//main.cc
#include <opencv2/core.hpp>
#include <iostream>
#include <string>
#include <glog/logging.h>
#include "yamlPasrser.h"

using namespace std;

int main(int argc, char* argv[])
{
   google::InitGoogleLogging(argv[0]); 

   FLAGS_log_dir = "./logs";

   std::string filename = "/home/zhipeng/catkin_ws/src/Kimera-VIO/params/Euroc/PipelineParams.yaml";

   YamlParser yaml_parser(filename);
   int backend_type;
   yaml_parser.getYamlParam("backend_type", &backend_type);

   std::cout << std::endl;
}

//yamlPasrser.h
#include <glog/logging.h>
#include <iostream>
#include <opencv2/core/core.hpp>



class YamlParser {
 public:

  YamlParser(const std::string& filepath) : fs_(), filepath_(filepath) {
    openFile(filepath, &fs_);
  }
  ~YamlParser() { closeFile(&fs_); }

 template <class T>
  void getYamlParam(const std::string& id, T* output) const {
    CHECK(!id.empty());
    const cv::FileNode& file_handle = fs_[id];
    CHECK_NE(file_handle.type(), cv::FileNode::NONE)
        << "Missing parameter: " << id.c_str()
        << " in file: " << filepath_.c_str();
    file_handle >> *CHECK_NOTNULL(output);
  }
  
 private:
  cv::FileStorage fs_;
  std::string filepath_;

 private:
  void openFile(const std::string& filepath, cv::FileStorage* fs) const {

    CHECK(!filepath.empty()) << "Empty filepath!";
    try {
      CHECK_NOTNULL(fs)->open(filepath, cv::FileStorage::READ);
    } catch (cv::Exception& e) {
      LOG(FATAL) << "Cannot open file: " << filepath << '\n'
                 << "OpenCV error code: " << e.msg;
    }
    LOG_IF(FATAL, !fs->isOpened())
        << "Cannot open file in parseYAML: " << filepath
        << " (remember that the first line should be: %YAML:1.0)";
  }

  inline void closeFile(cv::FileStorage* fs) const {
    CHECK_NOTNULL(fs)->release();
  }

};

that's weired. why Kimera-VIO just could not read .yaml file ?