Robotics StackExchange | Archived questions

pcl cuda cuda_runtime.h no such file or directory with ros

Hello everybody.

I'm trying to use cuda in my pcl/ros application. I want to use the bodyparts_detector (pcl/gpu/people).

If I compile my project with the include:

#include <pcl/gpu/people/bodyparts_detector.h>

it throws everytime the error:

/usr/local/include/pcl-1.8/pcl/gpu/people/labelcommon.h:44:71: fatal error: cudaruntime.h: No such file or directory
compilation terminated.
How can I solve this issue?

My CMakeLists.txt looks like this:

cmakeminimumrequired(VERSION 2.8.3) add_definitions(-std=c++11) project(project)

findpackage(catkin REQUIRED COMPONENTS cvbridge cmakemodules imagetransport pclconversions pclros roscpp messageandservice ) findpackage(OpenCV REQUIRED) findpackage(PCL REQUIRED) find_package(CUDA REQUIRED)

adddefinitions(${PCLDEFINITIONS})

Searching CUDA

FIND_PACKAGE(CUDA)

Include the FindCUDA script

INCLUDE(FindCUDA)

catkin_package(

)

addexecutable(${PROJECTNAME} src/main.cpp)

targetlinklibraries(${PROJECTNAME} ${catkinLIBRARIES} ${OpenCVLIBS} ${PCLLIBS} ${PCL_LIBRARIES} )

(This file is just a bit cleaned up now)

I hope someone can help me :)

Thank, Tobi

Asked by tobi93 on 2017-09-05 10:40:43 UTC

Comments

I have similar problem, but I have another problem:

nvcc fatal : A single input file is required for a non-link phase when an outputfile is specified

CMake Error at XXXXX_generated_XXXXX.cu.o.cmake:207 (message):

So I want to ask you, how to pass compile?

Asked by lgLeo on 2018-01-31 10:10:37 UTC

Answers