Segmentation fault (core dumped)

[Fedioo]

the Function pcl::fromROSMsg in lidar_detector/src/lib/node_lib.cpp is showing that error. any help please!

[RonaldEnsing]

It is difficult to look into this without having the exact steps of how to reproduce.

[DevonMorris]

I dug into this a bit today with gdb. It comes down to the pattern variable not having size 4 after processCircles. Here is the backtrace.

#0  index_eigen_matrix2D (in=..., indices=...) at /ws/src/common/src/robust_least_squares.cpp:258
#1  0x00007f2e22a1dc37 in findCorrespondences2D (in=..., out=..., all_permutations=...) at /ws/src/common/src/robust_least_squares.cpp:302
#2  0x00007f2e22a1e400 in refine2D (calibration_board=..., detections_3d=..., threshold_inlier=threshold_inlier@entry=0.019999999552965164)
    at /ws/src/common/src/robust_least_squares.cpp:392
#3  0x00007f2e28c3c793 in lidar_detector::refinement (pattern=..., config=...) at /ws/src/lidar_detector/src/lib/keypoint_detection.cpp:366
#4  0x00007f2e28c41481 in lidar_detector::keypointDetection (in=..., config=...) at /ws/src/lidar_detector/src/lib/keypoint_detection.cpp:401
#5  0x00007f2e2b1328e5 in lidar_detector::LidarDetectorNode::callback (this=0x7ffee674ef80, in=...) at /ws/src/lidar_detector/src/lib/node_lib.cpp:66
#6  0x00007f2e2b13940e in boost::function1<void, boost::shared_ptr<sensor_msgs::PointCloud2_<std::allocator<void> > const> const&>::operator() (a0=...,
    this=<optimized out>) at /usr/include/boost/function/function_template.hpp:759
#7  boost::detail::function::void_function_obj_invoker1<boost::function<void (boost::shared_ptr<sensor_msgs::PointCloud2_<std::allocator<void> > const> const&)>, void, boost::shared_ptr<sensor_msgs::PointCloud2_<std::allocator<void> > const> >::invoke(boost::detail::function::function_buffer&, boost::shared_ptr<sensor_msgs::PointCloud2_<std::allocator<void> > const>) (function_obj_ptr=..., a0=...) at /usr/include/boost/function/function_template.hpp:159
#8  0x00007f2e2b13fc76 in boost::function1<void, boost::shared_ptr<sensor_msgs::PointCloud2_<std::allocator<void> > const> >::operator() (a0=..., this=0x55e04f8e6fe8)
    at /usr/include/boost/function/function_template.hpp:759
#9  ros::SubscriptionCallbackHelperT<boost::shared_ptr<sensor_msgs::PointCloud2_<std::allocator<void> > const> const&, void>::call (this=0x55e04f8e6fe0, params=...)
    at /opt/ros/melodic/include/ros/subscription_callback_helper.h:144
#10 0x00007f2e29ba4e92 in ros::SubscriptionQueue::call() () from /opt/ros/melodic/lib/libroscpp.so
#11 0x00007f2e29b4f559 in ros::CallbackQueue::callOneCB(ros::CallbackQueue::TLS*) () from /opt/ros/melodic/lib/libroscpp.so
#12 0x00007f2e29b512fb in ros::CallbackQueue::callAvailable(ros::WallDuration) () from /opt/ros/melodic/lib/libroscpp.so
#13 0x00007f2e29ba8a39 in ros::SingleThreadedSpinner::spin(ros::CallbackQueue*) () from /opt/ros/melodic/lib/libroscpp.so
#14 0x00007f2e29b9137b in ros::spin() () from /opt/ros/melodic/lib/libroscpp.so
#15 0x000055e04dc07e17 in main (argc=<optimized out>, argv=<optimized out>) at /ws/src/lidar_detector/src/node.cpp:29

In frame 3, you can see that pattern is empty

(gdb) p pattern
$13 = {_vptr.PointCloud = 0x7f2e2b347400 <vtable for pcl::PointCloud<pcl::PointXYZ>+16>, header = {seq = 0, stamp = 0, frame_id = ""},
  points = std::vector of length 0, capacity 0, width = 0, height = 0, is_dense = true,
  sensor_origin_ = {<Eigen::PlainObjectBase<Eigen::Matrix<float, 4, 1, 0, 4, 1> >> = {<Eigen::MatrixBase<Eigen::Matrix<float, 4, 1, 0, 4, 1> >> = {<Eigen::DenseBase<Eigen::Matrix<float, 4, 1, 0, 4, 1> >> = {<Eigen::DenseCoeffsBase<Eigen::Matrix<float, 4, 1, 0, 4, 1>, 3>> = {<Eigen::DenseCoeffsBase<Eigen::Matrix<float, 4, 1, 0, 4, 1>, 1>> = {<Eigen::DenseCoeffsBase<Eigen::Matrix<float, 4, 1, 0, 4, 1>, 0>> = {<Eigen::EigenBase<Eigen::Matrix<float, 4, 1, 0, 4, 1> >> = {<No data fields>}, <No data fields>}, <No data fields>}, <No data fields>}, <No data fields>}, <No data fields>}, m_storage = {m_data = {array = {0, 0, 0, 0}}}}, <No data fields>},
  sensor_orientation_ = {<Eigen::QuaternionBase<Eigen::Quaternion<float, 0> >> = {<Eigen::RotationBase<Eigen::Quaternion<float, 0>, 3>> = {<No data fields>}, <No data fields>},
    m_coeffs = {<Eigen::PlainObjectBase<Eigen::Matrix<float, 4, 1, 0, 4, 1> >> = {<Eigen::MatrixBase<Eigen::Matrix<float, 4, 1, 0, 4, 1> >> = {<Eigen::DenseBase<Eigen::Matrix<float, 4, 1, 0, 4, 1> >> = {<Eigen::DenseCoeffsBase<Eigen::Matrix<float, 4, 1, 0, 4, 1>, 3>> = {<Eigen::DenseCoeffsBase<Eigen::Matrix<float, 4, 1, 0, 4, 1>, 1>> = {<Eigen::DenseCoeffsBase<Eigen::Matrix<float, 4, 1, 0, 4, 1>, 0>> = {<Eigen::EigenBase<Eigen::Matrix<float, 4, 1, 0, 4, 1> >> = {<No data fields>}, <No data fields>}, <No data fields>}, <No data fields>}, <No data fields>}, <No data fields>}, m_storage = {m_data = {array = {0, 0, 0, 1}}}}, <No data fields>}}, mapping_ = {px = 0x0, pn = {
      pi_ = 0x0}}}

I believe the cause of this is due to processCircles having no error checking to ensure at least 4 points were found https://github.com/tudelft-iv/multi_sensor_calibration/blob/master/lidar_detector/src/lib/keypoint_detection.cpp#L278