libmars_core.so: cannot open shared object file

[electricmanlpl]

Thanks for sharing, but when I run, it outputs error: ./example_app: error while loading shared libraries: libmars_core.so: cannot open shared object file: No such file or directory

[ygling2008]

Thanks for sharing too. I meet the same error. I have checked all the uploaded files in this GitHub. It seems that the authors have not uploaded this limars_core.so. Hope they will solve this issue soon.

[dawnos]

They have uploaded, but x86 version. That will leads to the problem on x64 machine. I have opened an issue on the request for x64 version.

[paulmrinal]

The libmars_core.so is available in the 'core' folder. Can you please mention on what machine and OS you are trying to run it. The desktop version runs on 64bit intel machines with Ubuntu 14.04

[electricmanlpl]

@paulmrinal thanks for your reply. I am running on AMD64 machine with Ubuntu 14.04

[ygling2008]

I run it on the 64bit machine with Ubuntu 14.04, and meet the same error (error while loading shared libraries: libmars_core.so).

[paulmrinal]

Can you please try adding the "build" directory to your LD_LIBRARY_PATH? export LD_LIBRARY_PATH="path-to-build-directory":$LD_LIBRARY_PATH

[dawnos]

Thank you! It works. But the next problem is, when I ran the program, I tells "KEYimu_dt*NOT FOUND". What may be the problem?

[paulmrinal]

We uploaded a sample config file in the config folder, please use this one. The imu_dt is the delta_time between the imu samples.

[ygling2008]

Many thanks to @paulmrinal , it works now. Great Job!

[dawnos]

It works! Thank you! @paulmrinal

[ygling2008]

@paulmrinal It seems that this lib can only run for 10s. I can not run the whole testing sequence provided by your website. Are there any ways to run for longer time?

[paulmrinal]

In the sample app there is a variable named "image_num", you need to set it properly. It was just set to 500, that's why it was not running for the full sequence. Here is a sample code snippet that you can use to get the "image_num":

#include <glob.h>
#include <string>

----------------------------------------------------------------

std::string path(left_image_path);
glob_t image_gl;
size_t image_num = 0;
std::string pgm_path = path + "*.pgm";
if( glob(pgm_path.c_str(), GLOB_NOSORT, NULL, &image_gl) == 0){
  image_num = image_gl.gl_pathc;
}
globfree(&image_gl);

[ygling2008]

@paulmrinal Thanks a lot. Is there an extra mode (MONO) using only left images and IMU measurements?

[paulmrinal]

The current binary does not support mono mode, we are planning to soon update the binaries to give options for mono and also open some more config parameters, so that people can configure the system for particular environments and motion profile.

[ygling2008]

Thanks @paulmrinal in advance. It is a great opportunity for the readers in the SLAM society to compare different algorithms, especially compare their results with the ones from your group (that is well-known for VINS).