Skip to content

alex-luxonis/test-depthai_dual_cam_ros_detector

 
 

Repository files navigation

depthai_dual_cam_ros_detector

This repo provides the code and instructions to reproduce an issue I encounter while running a depthai pipeline of two color cameras and detection networks with OAK-FFC-3P. The issue happens specifically when connecting with two IMX577 modules from Arducam.

Hardware

OAK-FFC-3P configuration

As comparison, we also ran the same pipeline with IMX214 from Luxonis. Here are the two configuration.

  • 2-IMX577: OAK-FFC-3P with two IMX577 modules on CAM_B and CAM_C

2 IMX577

  • 2-IMX214: OAK-FFC-3P with two IMX214 modules on CAM_B and CAM_C

2 IMX214

The hosts

All the tests have been run with two different hosts seperately for comparison.

  • workstation host: a laptop running Ubuntu 22.04.3
  • pi host: a Raspberry Pi 4 Model B with precompiled Depthai OS (OAK_CM4_POE_V10_64bit, Debian GNU/Linux 12 bookworm).

Reproduce the issue

I built a docker image with depthai-ros noetic and our minimum reproducible application. A run script run.sh will launch the pipeline and use pipeline_graph to show the pipeline details.

Steps

  • Clone this repo by git clone https://github.com/lincolnxlw/depthai_dual_cam_ros_detector.git.
  • Download my prebuilt docker image by docker pull lincolnxlw/my-depthai-ros.
  • Run the main script in the repo's root folder ./run.sh.

Expected behavior

Both camera nodes should return expected 15 FPS.

Expected FPS

Observed behavior

With configuration 2-IMX577, one camera will return 0 FPS and never recover during run.

Failure FPS

Detailed test results

workstation host

  • 2-IMX577: failed 9 times out of 30 times
  • 2-IMX214: failed 0 times out of 30 times

pi host

  • 2-IMX577: failed 11 times out of 30 times
  • 2-IMX214: failed 0 times out of 30 times

Other details

All connections during the testing are tripple checked. I also randomly paired up IMX577 modules and the result persist.

Detail 1 Detail 2 Detail 3 All Tested Modules

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • C++ 83.6%
  • Shell 6.5%
  • Dockerfile 6.0%
  • CMake 3.9%