Open Greg8978 opened 3 years ago
Greg8978
commented
3 years ago If anyone bump into the same issue, it comes from this [commit]( https://github.com/BerkeleyAutomation/perception/commit/1dd48f392bc3777b2111636180cdac31d620751d):
it removes CameraInstrinsics!
Reverting perception repo fixed it for me. I had other issues though, mainly compatibility between tensorflow1 and python3.8. I ended up installing only gqcnn with python 3.7, which may be enough for me.
Jasonleeo-cpu
commented
3 years ago Could you help me to solve the problem as follows or tell me some details about it ? Unable to import meshrender shared library! Rendering will not work. Likely due to missing Boost.Numpy Boost.Numpy can be installed following the instructions in https://github.com/ndarray/Boost.NumPy Thanks in advance!!
Greg8978
commented
3 years ago Hey! Depending on what version you are using, you should try this: https://github.com/BerkeleyAutomation/meshpy/issues/1
Here is what I did:
#include <boost/python.hpp>
#include "boost/python/extract.hpp"
//#include "boost/python/numeric.hpp"
#if BOOST_VERSION < 106500
#include <boost/python/numeric.hpp>
#else
#include <boost/python/numpy.hpp>
#endif
//#include <boost/numpy.hpp>
...
Then in the code, replace boost::python::numeric by
boost::python::numpy
boost::python::tuple render_mesh(boost::python::list proj_matrices,
unsigned int im_height,
unsigned int im_width,
boost::python::numpy::ndarray verts,
boost::python::numpy::ndarray tris,
boost::python::numpy::ndarray norms,
boost::python::numpy::ndarray mat_props,
boost::python::numpy::ndarray light_props,
bool enable_lighting = false,
bool debug = false)(...)
// append ndarray color image to list
boost::python::tuple color_shape = boost::python::make_tuple(im_height, im_width, 3);
boost::python::numpy::dtype color_dt = boost::python::numpy::dtype::get_builtin<unsigned char>();
boost::python::numpy::ndarray color_arr = boost::python::numpy::from_data(color_result, color_dt, color_shape,
boost::python::make_tuple(color_shape[1]*color_shape[2]*sizeof(unsigned char),
color_shape[2]*sizeof(unsigned char),
sizeof(unsigned char)),
boost::python::object());
color_ims.append(color_arr.copy());
// append ndarray depth image to list
boost::python::tuple depth_shape = boost::python::make_tuple(im_height, im_width);
boost::python::numpy::dtype depth_dt = boost::python::numpy::dtype::get_builtin<float>();
boost::python::numpy::ndarray depth_arr = boost::python::numpy::from_data(depth_result, depth_dt, depth_shape,
boost::python::make_tuple(depth_shape[1]*sizeof(float),
sizeof(float)),
boost::python::object());
depth_ims.append(depth_arr.copy());
}(...)
// Test function for multiplying an array by a scalar
boost::python::list mul_array(boost::python::numpy::ndarray data, int x)
{
// Access a built-in type (an array)
boost::python::numpy::ndarray a = data;
long int bufLen;
void const *buffer;(...)
const boost::python::tuple& shape = boost::python::extract<boost::python::tuple>(a.attr("shape"));
std::cout << "Shape " << boost::python::extract<int>(shape[0]) << " " << boost::python::extract<int>(shape[1]) << std::endl;
boost::python::numpy::dtype dt = boost::python::numpy::dtype::get_builtin<double>();
boost::python::numpy::ndarray result = boost::python::numpy::from_data(mult, dt, shape,
boost::python::make_tuple(shape[0]*sizeof(double),
sizeof(double)),
boost::python::object());
(...)
// Expose classes and methods to Python
BOOST_PYTHON_MODULE(meshrender) {
Py_Initialize();
boost::python::numpy::initialize();
def("mul_array", &mul_array);
def("render_mesh", &render_mesh);
}
Ultraopxt
commented
2 years ago Hi! I know it's not a tested version for ubuntu, but I'm still commited to make this work on Noetic / ubuntu 20.04
I ran into several issues that I could solve with various documentation / issues already posted. I finally succeeded, with some fix here and there, to compile without error. My issue is when I launch the test, I get the following result:
running test WARNING: Testing via this command is deprecated and will be removed in a future version. Users looking for a generic test entry point independent of test runner are encouraged to use tox. running egg_info creating src/dex_net.egg-info writing src/dex_net.egg-info/PKG-INFO writing dependency_links to src/dex_net.egg-info/dependency_links.txt writing requirements to src/dex_net.egg-info/requires.txt writing top-level names to src/dex_net.egg-info/top_level.txt writing manifest file 'src/dex_net.egg-info/SOURCES.txt' reading manifest file 'src/dex_net.egg-info/SOURCES.txt' writing manifest file 'src/dex_net.egg-info/SOURCES.txt' running build_ext database_test (unittest.loader._FailedTest) ... ERROR grasping_test (unittest.loader._FailedTest) ... ERROR learning_test (unittest.loader._FailedTest) ... ERROR ====================================================================== ERROR: database_test (unittest.loader._FailedTest) ---------------------------------------------------------------------- ImportError: Failed to import test module: database_test Traceback (most recent call last): File "/usr/lib/python3.8/unittest/loader.py", line 154, in loadTestsFromName module = __import__(module_name) File "/home/user_adm/catkin_ws/src/dex-net/test/database_test.py", line 37, in <module> from perception import CameraIntrinsics, RenderMode ImportError: cannot import name 'CameraIntrinsics' from 'perception' (/home/user_adm/catkin_ws/devel/lib/python3/dist-packages/perception/__init__.py) ====================================================================== ERROR: grasping_test (unittest.loader._FailedTest) ---------------------------------------------------------------------- ImportError: Failed to import test module: grasping_test Traceback (most recent call last): File "/usr/lib/python3.8/unittest/loader.py", line 154, in loadTestsFromName module = __import__(module_name) File "/home/user_adm/catkin_ws/src/dex-net/test/grasping_test.py", line 36, in <module> from perception import CameraIntrinsics ImportError: cannot import name 'CameraIntrinsics' from 'perception' (/home/user_adm/catkin_ws/devel/lib/python3/dist-packages/perception/__init__.py) ====================================================================== ERROR: learning_test (unittest.loader._FailedTest) ---------------------------------------------------------------------- ImportError: Failed to import test module: learning_test Traceback (most recent call last): File "/usr/lib/python3.8/unittest/loader.py", line 154, in loadTestsFromName module = __import__(module_name) File "/home/user_adm/catkin_ws/src/dex-net/test/learning_test.py", line 38, in <module> from dexnet.learning import RandomBinaryObjective, RandomContinuousObjective, UniformAllocationMean, ThompsonSampling, GaussianUniformAllocationMean, MaxIterTerminationCondition File "/home/user_adm/catkin_ws/src/dex-net/src/dexnet/__init__.py", line 23, in <module> from abstractstatic import abstractstatic ModuleNotFoundError: No module named 'abstractstatic' ---------------------------------------------------------------------- Ran 3 tests in 0.000s FAILED (errors=3) Test failed: <unittest.runner.TextTestResult run=3 errors=3 failures=0> error: Test failed: <unittest.runner.TextTestResult run=3 errors=3 failures=0>I installed perception and everything required following this document.
Any help would be greatly appreciated :)
Hi, Did you run dexnet on ubuntu20.04 succesfully?
GregoireMilliez
commented
2 years ago Hi!
We ran only with GQCNN, we gave up on Dex-Net. Even GQCNN was hard to use as the normalization of data is complex without Dex-Net tools.
However we found this document that may be of interest for you: https://www.imperial.ac.uk/media/imperial-college/faculty-of-engineering/computing/public/1819-ug-projects/SaporaS-Grasp-Quality-Deep-Neural-Networks-for-Robotic-Object-Grasping.pdf
Cheers
Ultraopxt
commented
2 years ago Hi!
We ran only with GQCNN, we gave up on Dex-Net. Even GQCNN was hard to use as the normalization of data is complex without Dex-Net tools.
However we found this document that may be of interest for you: https://www.imperial.ac.uk/media/imperial-college/faculty-of-engineering/computing/public/1819-ug-projects/SaporaS-Grasp-Quality-Deep-Neural-Networks-for-Robotic-Object-Grasping.pdf
Cheers
Thanks very much!
Could you please share the github link of this paper?
GregoireMilliez
commented
2 years ago 
elevenjiang1
commented
2 years ago I also have same problem! I think the main problem is the version are too mess in autolab github !
elevenjiang1
commented
2 years ago It seems that in Python3,or in high version in autolab_core,from perception import CameraIntrinsics need to change to from autolab_core import CameraIntrinsics
I have setup dexnet on Ubuntu 16.04 python2, but give up setup python3 in Ubuntu18.04.Also, I have setup gqcnn on Ubuntu16.03 in python3. All the environment setup is in docker. If you need help, I can give you more detail~
weiyunpei
commented
2 years ago It seems that in Python3,or in high version in autolab_core,
from perception import CameraIntrinsicsneed to change tofrom autolab_core import CameraIntrinsicsI have setup dexnet on Ubuntu 16.04 python2, but give up setup python3 in Ubuntu18.04.Also, I have setup gqcnn on Ubuntu16.03 in python3. All the environment setup is in docker. If you need help, I can give you more detail~
hello,could you please send a github link , I am doing my undergraduate project.THANKS!
elevenjiang1
commented
2 years ago It seems that in Python3,or in high version in autolab_core,
from perception import CameraIntrinsicsneed to change tofrom autolab_core import CameraIntrinsicsI have setup dexnet on Ubuntu 16.04 python2, but give up setup python3 in Ubuntu18.04.Also, I have setup gqcnn on Ubuntu16.03 in python3. All the environment setup is in docker. If you need help, I can give you more detail~hello,could you please send a github link , I am doing my undergraduate project.THANKS!
Sorry, I haven't cleaned my code now. If you need more help, maybe you can ask them by email: 1206413225@qq.com
Manu752
commented
2 months ago Hi , Were you guys able to get dex-net /GPD eunning on ubuntu 20.4 with ros noetic ? ...I have a project underway and I cant seem to get past the build stage (step 4) of the repo :(
Greg8978
commented
2 months ago Hi, on our side we actually gave up on DexNet, we moved to RGBMatter rgb_matters and suctionNet. It was easier to use. https://github.com/GouMinghao/rgb_matters
Manu752
commented
2 months ago thank you for the prompt response :) ... does RGBMatter rgb_matters work with move it ? I went threw the repo but couldn't find any doc on how to integrate it with moveit .
Greg8978
commented
2 months ago Yes, we deployed it on Niryo Ned 2, which is using MoveIt. GraspNet will give you the gripping pose. You will then need to feed it to MoveIt
Hi! I know it's not a tested version for ubuntu, but I'm still commited to make this work on Noetic / ubuntu 20.04
I ran into several issues that I could solve with various documentation / issues already posted. I finally succeeded, with some fix here and there, to compile without error. My issue is when I launch the test, I get the following result:
I installed perception and everything required following this document.
Any help would be greatly appreciated :)