I created code based of the tensorRT sample sampleFasterRCNN and I am unable to compile it for ARM aarch64, which is the format the samples on the DrivePX2 for tensorRT are in. I can only successfully compile for x86-64 on the host; if I set TARGET to “aarch64” when I use make, I get errors that certain files for compilation cannot be located. I have tried compiling on the DrivePX2 in that format, but to no avail (I get errors that it can’t find files the host can). The documentation for tensorRT does not seem to explain how to cross-compile for different targets. I am unsure what the procedure for cross-compiling a tensorRT sample to get an executable for ARM aarch64 to run on the Drive PX2 is.
Could you check compiling the sample on board directly. It is similar to how to you include a sample on x86.