Hey everyone, I’m trying to get the Arducam ToF camera working on my Jetson Nano. I followed the official getting-started guide at the Arducam docs site, but when I run ./compile.sh it fails during the build. The error seems to be about a missing library — something to do with Open3D / libopen3d for the pointcloud functionality. Has anyone else run into this?
Welcome! Can you share a bit more about your environment? Which JetPack / SDK version are you on, and what Python version is your Nano running? That’ll help narrow things down.
Sure thing. I’m on a Jetson Nano with JetPack SDK 32.7.6 (tegra 4.9), and it’s running Python 3.6. I actually tried installing the Open3D Python package with pip3 to satisfy the dependency, but that failed too — Open3D requires Python ≥ 3.8 and the Nano is stuck on 3.6. So I’m kind of in a dead end there.
Right, that Python version wall on the Nano is a common pain point with Open3D. One workaround is to skip the Python package and install the system-level C++ development library instead. That should give the compile script the headers and libs it needs without going through pip. Try this:
sudo apt install libopen3d-dev
Then re-run ./compile.sh and let me know how it goes.
Gave that a shot — sudo apt install libopen3d-dev went through fine, but unfortunately ./compile.sh still fails with the same kind of error. It doesn’t seem to have resolved the missing dependency. I also tried poking around manually. The compile script seems to rely on an ArducamDepth header, so I attempted to install that separately. Tried pip3 install arducamdepth and got an error — something like “pytron can’t install header.” Honestly not sure what that means. Feels like I’m chasing dependencies in circles here.
That’s frustrating. Just to rule out other environment issues — do you have OpenCV installed and working on that Nano? The ToF examples lean on it pretty heavily, so it’s worth confirming that’s in good shape before we dig deeper into the Arducam-specific stuff.
Yeah, OpenCV is installed — version 4.9.0. I’ve run a few of the standard OpenCV examples and they work without any issues, so I don’t think that’s the culprit.
Okay, good to know OpenCV is solid. I suspect the official repo’s dependency handling might not play nicely with the Jetson Nano’s older environment. There’s an alternative fork that’s been adapted to handle these kinds of dependency quirks better on Jetson platforms. Give this a try:
git clone https://github.com/antonio-ard/Arducam_tof_camera.git
Then follow the build instructions in that repo’s README at GitHub - antonio-ard/Arducam_tof_camera · GitHub. It may handle the Open3D linkage differently and should be more forgiving on Python 3.6. Let us know how it goes!
Thanks — I’ll clone that repo and run through the build steps there. Will report back once I’ve had a chance to try it out. Appreciate the help so far.