Instability with USB cameras in vehicles
We are using 2 x USB 2.0 RGB cameras to capture image at 20fps with a mini-ITX motherboard from ASUS. In the mini-ITX we are running a linux system with driver Video4Linux, and open CV for capturing and processing The system is intended for running in trucks and buses.
For connecting the cameras, We are using 3.5m UTP cable with RJ45 instead of 100 ohms USB cable and connector, and a propietary adapter board to convert female pin and enter to the internal USB headers of the motherboard, so our channel has a slight desadaptation. The UTP cable twisted pair has 90ohms diff characteristic impedance instead of the official 100ohms that we find in the USB twisted pair, as stated by USB.org
We have an electronic signal in the propietary adapter board in each USB channel in which we can bring down and up a camera disabling USB 5VDC and enabling again.
We test our equipment in laboratory and works fine, we can open both cameras and work with them without any problems for days. If we issue a manual reset of the 5VDC line of the camera, we can open it again without problems.
Our issue: When we install our system in vehicles, sometimes it works as stable as in laboratory, but in some cases the cameras will hang, and when the equipment tries reseting the 5VDC line, it's impossible to bring them back again.
I know that our first basic problem is not using a proper USB 100ohms channel, but I would like to know any thoughs about the case, people who has experience working with V4L and usb cameras.
Thanks very much! Sebastian
And the connection with Opencv is...???
...and as you asked for my thoughts: I don't think it's a good idea to play with the cables like this; use the provided cables unless you really know what you're doing.
For difficult environments (like moving vehicles) I don't recommend at all using webcams. Rather go with robust industrial cameras, like the Basler Ace, IDS uEye, Point Grey etc.