I’m setting up a NVIDIA Jetson Orin NX with JetPack 6.2.1, L4T r36.4.0.
I have a cpp program that runs on libtorch, onnxruntime, pangolin, and other libraries, that runs on a docker container in my host computer (a regular 22.04 ubuntu).
When creating a new docker file with all the libraries for the Jetson, I’m having trouble with compiling the code, despite using the same project and modifying paths accordingly on the CMakelists.txt. The main issues are regarding libtorch in the arm architecture.
I provide both original Dockerfile that works correctly on the Ubuntu and the jetson-adapted Dockerfile, and would love any suggestions on what to include or exclude so I can get the exact behaviour I have on the ubuntu, but in the Jetson.