So I cloned a git repository which is registrating point clouds using probabilistic methods such as GMMs (Gaussian-Mixture-Models) but also incorporates Open3D. Because the registration is running on the GPU instead of the CPU I needed to install CUDA. Mainly because the repository imports the CuPy-Package which in turn also requires CUDA. Because I had some problems installing and building the latest versions of CuPy, I solved it by just using an older version, in my case version 10.2, which seems to run on my computer and on my GPU (it's a GTX 1050). After that CuPy was successfully imported into the project. Installation and import of Open3D worked also fine. However, after running a python test-file which is registering two point clouds, I got a weird error message in the terminal, which also crashed the program. It says Module 'Open3D' has no attribute 'read_point_cloud'
The corressponding code line is this: source = o3.read_point_cloud(filepath) Now, I made some research by reading the Open3D documentation and I found out that read_point_cloud is an already provided default method in the Open3D library, in older versions, as well as in its current version 0.18.0. It is also the same version which I installed via pip previously. Could it somehow be related to the CUDA problem I described above or am I missing something here?