example simulator_3d_line get "Segmentation fault (core dumped)"

[maShimizu]

Hi here

I get "Segmentation fault (core dumped)" using simulator_3d_line example code. log: Using CSparse poseDim -1 landMarkDim -1 blockordering 0Creating simulator

Creating robot Creating line sensor Creating landmark line 1 Creating landmark line 2 Creating landmark line 3 Segmentation fault (core dumped)

And, I try 3d slam and run original code. but sometimes get "Segmentation fault" same as above, too. I have suspicious about "g2o::ParameterSE3Offset", because I just try simple code as below and get "segmentation fault". Please anyone let me know where I should modify this code or check.

Code( a port of ):

include "g2o/core/sparse_optimizer.h"

include "g2o/core/block_solver.h"

include "g2o/core/linear_solver.h"

include "g2o/core/factory.h"

include "g2o/core/optimization_algorithm_factory.h"

include "g2o/core/optimization_algorithm_gauss_newton.h"

include "g2o/core/optimization_algorithm_levenberg.h"

include "g2o/solvers/csparse/linear_solver_csparse.h"

include "g2o/solvers/eigen/linear_solver_eigen.h"

include "g2o/types/slam3d/types_slam3d.h"

void nviewTracker::BundleAdjustment( std::vector<Frame> pframes ) { g2o::SparseOptimizer optimizer; optimizer.setVerbose(false);

std::unique_ptr<g2o::BlockSolver_6_3::LinearSolverType> linearSolver;
linearSolver = g2o::make_unique<g2o::LinearSolverCSparse<g2o::BlockSolver_6_3::PoseMatrixType>>();

g2o::OptimizationAlgorithmLevenberg* solver = new g2o::OptimizationAlgorithmLevenberg( g2o::make_unique<g2o::BlockSolver_6_3>(std::move(linearSolver)) );
optimizer.setAlgorithm(solver);

g2o::ParameterSE3Offset* odomOffset = new g2o::ParameterSE3Offset;
//odomOffset->setId(0);
//optimizer.addParameter(odomOffset);

}

[maShimizu]

Hi all

I've already posted above, and I try to resolve now. I run g2o code on AWS ubuntu. AWS instance don't have GPU, I install and use OpenGL/mesa. I guess OpenGL/mesa have any trouble.

Does anyone try g2o on AWS?

[01001HR]

I guess there were many bugs after merging "Make G2O Compatible With Hardware That Only Supports Single Precision Floating Point Types", try checkout commit before that.

[jumonatr]

I don't think that check-in has anything to do with it, as it just adds a using definition for changing the default number type. It also doesn't change anything by default, you have to opt-in to change the type.

Have you tried attaching a debugger to it to figure out what the callstack is?

[maShimizu]

I don't change types, I use just original slam_3d example types. Anyway, I try a debugger. But, I can't find attaching a debugger in your comment. or is a debugger in g2o?

[RainerKuemmerle]

http://senarvi.github.io/2016/07/05/stack-trace-with-gdb

[RainerKuemmerle]

regarding OpenGL. There is a build config variable to exclude OpenGL. Set G2O_USE_OPENGL to false in your cmake configuration.