Resolving problems for installing HDE on a new MAC machine about human-dynamics-estimation HOT 12 CLOSED

Just to understand, why are you using Ubuntu 16.04 ? The next releases of YARP will not support Ubuntu 16.04 (see https://github.com/robotology/yarp/blob/master/.github/CONTRIBUTING.md#supported-systems), and so except in cases where a given projects explicitly requires Ubuntu 16.04, we should aim to migrate to Ubuntu 18.04 .

from human-dynamics-estimation.

The iDynTree::ModelLoader class is contained in the iDynTree::idyntree-modelio-urdf exported target (see https://github.com/robotology/idyntree/blob/114671fb1a012f6214b35b8462d463fea99781fd/src/model_io/urdf/CMakeLists.txt#L57).

To be honest, I think it is wise to reccomended and document to just link against the list of exported targets ${iDynTree_LIBRARIES}, that contains all the targets exported by iDynTree, so you are sure to always get everything. The mapping between the specific iDynTree symbols and their target is not properly documented anywhere, and unfortunatly at the moment is not easy to deprecate a CMake imported target, see https://gitlab.kitware.com/cmake/cmake/issues/17824, so having the symbols <--> target mapping as part of the public API of the library complicates the process of refactoring the symbols between targets (i.e. moving a class for a target to another).

from human-dynamics-estimation.

@kouroshD please document also the details for setting up ma27 solver and other issues related to human model urdfs encountered today.

from human-dynamics-estimation.

• change /devices/IAnalogSesorToIWear/src/IAnalogSensorToIWear.cpp, becasue the current implementation is depended on yarp dev branch, by changing this lline it will be compatible with master branch:
  updated lines:

    bool getData(std::vector<double>& vector, const size_t offset = 0)
    {
-        std::copy(buffer.data() + offset, buffer.end() + offset, vector.data());
+        std::copy(buffer.data() + offset, buffer.data() + buffer.size() + offset, vector.data());
        return true;
    }

As discussed also with @diegoferigo, it may take long to get the release of the new YARP and for the moment it would be better to ensure the compatibility with the current master branch of YARP.

commet the line #add_subdirectory(HumanControlBoard) in file HDE/devices/CMakeLists.txt : it does not compile in macos currently.

We get the following error when trying to compile it in MacOS:

[ 42%] Linking CXX shared module ../../lib/yarp/HumanControlBoard.so
Undefined symbols for architecture x86_64:
  "iDynTree::ModelLoader::loadModelFromFile(std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char> > const&, std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char> > const&)", referenced from:
      hde::devices::HumanControlBoard::open(yarp::os::Searchable&) in HumanControlBoard.cpp.o
  "iDynTree::ModelLoader::model()", referenced from:
      hde::devices::HumanControlBoard::open(yarp::os::Searchable&) in HumanControlBoard.cpp.o
  "iDynTree::ModelLoader::isValid()", referenced from:
      hde::devices::HumanControlBoard::open(yarp::os::Searchable&) in HumanControlBoard.cpp.o
  "iDynTree::ModelLoader::ModelLoader()", referenced from:
      hde::devices::HumanControlBoard::open(yarp::os::Searchable&) in HumanControlBoard.cpp.o
  "iDynTree::ModelLoader::~ModelLoader()", referenced from:
      hde::devices::HumanControlBoard::open(yarp::os::Searchable&) in HumanControlBoard.cpp.o
ld: symbol(s) not found for architecture x86_64
clang: error: linker command failed with exit code 1 (use -v to see invocation)
make[2]: *** [lib/yarp/HumanControlBoard.so] Error 1
make[1]: *** [devices/HumanControlBoard/CMakeFiles/HumanControlBoard.dir/all] Error 2
make: *** [all] Error 2

from human-dynamics-estimation.

• change /devices/IAnalogSesorToIWear/src/IAnalogSensorToIWear.cpp, becasue the current implementation is depended on yarp dev branch, by changing this lline it will be compatible with master branch:
  updated lines:

    bool getData(std::vector<double>& vector, const size_t offset = 0)
    {
-        std::copy(buffer.data() + offset, buffer.end() + offset, vector.data());
+        std::copy(buffer.data() + offset, buffer.data() + buffer.size() + offset, vector.data());
        return true;
    }

As discussed also with @diegoferigo, it may take long to get the release of the new YARP and for the moment it would be better to ensure the compatibility with the current master branch of YARP.

This back-compatibility is introduced with https://github.com/Yeshasvitvs/wearables/pull/5/commits/8aa47b8b9d5b581ca1d64af7e6bd414864906fef. Soon to be merged

from human-dynamics-estimation.

Sure, I will add them following of this issue.

from human-dynamics-estimation.

It was an old installation, I didn't know about this fact. I will migrate to Ubuntu 18.04 soon for the ubuntu machine I am using.

from human-dynamics-estimation.

Great!

from human-dynamics-estimation.

iDynTree branch should be another branch, not master:

cd <idyntree>
git checkout -t feature/improvedRobotStatePublisher

build again the superbuild, but before change the configuartion to:

YCM_EP_DEVEL_MODE_iDynTree : ON

from human-dynamics-estimation.

• Add the install prefix for the wearable and HDE the path to the robotology superbuild install prefix.
• for the Wearable repo:
  • Clone it from this branch
  • change /devices/IAnalogSesorToIWear/src/IAnalogSensorToIWear.cpp, becasue the current implementation is depended on yarp dev branch, by changing this lline it will be compatible with master branch:
    updated lines:

    bool getData(std::vector<double>& vector, const size_t offset = 0)
    {
-        std::copy(buffer.data() + offset, buffer.end() + offset, vector.data());
+        std::copy(buffer.data() + offset, buffer.data() + buffer.size() + offset, vector.data());
        return true;
    }

commet the line #add_subdirectory(HumanControlBoard) in file HDE/devices/CMakeLists.txt : it does not compile in macos currently.

comment the follwoing lines in xml configuration file (Human.xml):

    <!--    <param name="useRPC">false</param>
        <param name="wearableRPCPorts">(/XSensSuit/WearableData/metadataRpc:o)</param> -->

from human-dynamics-estimation.

commet the line #add_subdirectory(HumanControlBoard) in file HDE/devices/CMakeLists.txt : it does not compile in macos currently.

We get the following error when trying to compile it in MacOS:

[ 42%] Linking CXX shared module ../../lib/yarp/HumanControlBoard.so
Undefined symbols for architecture x86_64:
  "iDynTree::ModelLoader::loadModelFromFile(std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char> > const&, std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char> > const&)", referenced from:
      hde::devices::HumanControlBoard::open(yarp::os::Searchable&) in HumanControlBoard.cpp.o
  "iDynTree::ModelLoader::model()", referenced from:
      hde::devices::HumanControlBoard::open(yarp::os::Searchable&) in HumanControlBoard.cpp.o
  "iDynTree::ModelLoader::isValid()", referenced from:
      hde::devices::HumanControlBoard::open(yarp::os::Searchable&) in HumanControlBoard.cpp.o
  "iDynTree::ModelLoader::ModelLoader()", referenced from:
      hde::devices::HumanControlBoard::open(yarp::os::Searchable&) in HumanControlBoard.cpp.o
  "iDynTree::ModelLoader::~ModelLoader()", referenced from:
      hde::devices::HumanControlBoard::open(yarp::os::Searchable&) in HumanControlBoard.cpp.o
ld: symbol(s) not found for architecture x86_64
clang: error: linker command failed with exit code 1 (use -v to see invocation)
make[2]: *** [lib/yarp/HumanControlBoard.so] Error 1
make[1]: *** [devices/HumanControlBoard/CMakeFiles/HumanControlBoard.dir/all] Error 2
make: *** [all] Error 2

Apparently, the problem here seems to be solved by adding iDynTree::idyntree-estimation to the target_link_libraries (in addition to iDynTree::idyntree-model) in order to use the ModelLoader in MacOS.
I don't think this is meant to be working in this way.
Am I correct @traversaro? In case we can open an issue in iDynTree.

(The device we are referring to is the humanControlBoard)

from human-dynamics-estimation.

Since I have not faced more problems, we can close this issue.

from human-dynamics-estimation.