Hi,
I’m working on a project with a Jetson Nano and have some questions about multi-camera and multi-display capabilities of the board.
A short overview of my project:
I’m working on a small project with a recumbent bike that already has a carbon hull, so it’s aerodynamic. The problem is that your view from inside the hull is limited, and so it can be dangerous to drive with it. That’s why I want to build some cameras (I think 4 are the best) around it to have a panoramic view. My first idea was to mount a 360° camera on the roof, but for my use case it’s not optimal. I also couldn’t find any good stitching software that runs fluently (with 15 fps or higher) on a Jetson Nano or Raspberry Pi. That’s why I don’t want to stitch the video streams anymore. If there’s any good option I overlooked, I’d try it. Another difficulty is, that the Jetson Nano only has two camera ports. That’s why I thought I can use two cameras via USB and two via the camera ports, but I don’t know how good that’ll work. I also found those Arducam camera adapters, but I don’t know how good they are and if there is a big disadvantage because of which I shouldn’t use them. I already found cameras like the PiCam360-CAMPT8MP that runs by USB. To display the videos, I want to use two small touch-dispays for Raspberry Pi. Of course, I won’t completely rely on the system. If it crashes, I’ll still be able to drive safely.
Summary:
I want to build a system for panoramic view with a Jetson Nano and four cameras.
My questions:
- Is there any method to stitch four video streams fluently (with more than 15 fps and with a Jetson Nano)?
- What is the best method to use four cameras (via USB and camera ports, or only USB…)?
- Can I connect the two touch-dispays with the Jetson Nano (the one via the HDMI port and the other by the Display Port)?
- Do you think I can “analyze” the four streams in real time with a program like OpenCV or TensorFlow, or is the Jetson Nano too slow for something like that?
- Do you have any ideas or tips how I can build the project as good as possible?
Many thanks in advance, Konrad.