I did some checking and it looks like replacing the camera might be a good way to go. I plugged in one of my old Logitech C310 cameras and measured the latency at 5-7 milliseconds at that resolution, which is acceptable. The latency I got the last time I did this sort of thing must have been coming from somewhere else (or it was misconfigured). That said, it was jello effecting pretty hard at those speeds.
I took robot for a spin and recorded some training data with that camera connected instead of the other one. All of my problems with the GPS vanished, so it looks like the original camera is the only thing causing the problem. I am ordering a similar wide-angle camera that is USB 2.0 from Amazon (arriving tomorrow). I still need to see if it will give acceptable performance. It seems like the more well known manufactures don't do 170 degree cameras. Thanks, Charlie On Tue, Apr 30, 2019 at 10:11 PM Charles West <[email protected]> wrote: > I would love to. The catch is that the camera came with a custom made > cable that directly hooks into it. The USB3 capability gives really low > latency (<5 milliseconds if I remember right) and it has a really wide > angle (70 degrees?). I also can't switch cameras without causing > difficulties using my existing training data. > > That said, I've only got about 2 days of proper data so far with this > setup so an alternate camera could be somewhat possible. Research so far > does agree that it is USB3 causing the interference. I was trying to do > USB because it makes it really easy to port between platforms, but I notice > that this particular camera module gives me roughly similar speeds over > USB2 legacy ports. Does anyone know of a good wide angle USB camera (60+ > degrees) that can deliver 224x224 with low latency (<5-10 millisecond)? > > Also, thank you Kevin & Pete. > > On Tue, Apr 30, 2019 at 8:14 PM Pete Soper via TriEmbed < > [email protected]> wrote: > >> The aluminum foil should be replaced with a good USB cable that has the >> "bulges" indicating RF chokes. Adding additional RF chokes to the cable on >> the end closest to the camera may help further. The GPS cable should >> probably get an RF choke or two too. My final suggestion is to hit the >> camera with an eight pound sledge and buy a different one that has an FCC >> label. >> -Pete >> >> On 4/30/19 10:33 AM, Charles West via TriEmbed wrote: >> >> Hello from the North! >> >> I'm still working on the golf course robot. Recently I've run into a >> rather odd issue. When I put everything together for my 1/8 scale RC car >> platform, my GPS module refused to converge. I spent a solid 4 days >> debugging it only to eventually discover that the problem appeared to be >> that my USB 3 camera was putting out large amounts of RF interference of a >> spectrum suitable to interfere with my GPS even when there was a metal >> ground plane between the GPS and the rest of the equipment. >> >> Eventually, a combination of wrapping the camera in metal tape, placing >> the GPS with a ground plane elevated way above the rest of the robot and >> wrapping the camera's cable in aluminum foil finally got the GPS to >> converge. The only problem is, the accuracy of the GPS is far worse than >> in the previous version of the robot. Originally, I got within a 2-3 >> meters with an occasional drift. Now it squiggles all over the place. >> >> Does anyone know what I should try next to get decent performance and/or >> hunt down any interference for the GPS? >> >> Thanks, >> Charlie West >> >> _______________________________________________ >> Triangle, NC Embedded Computing mailing list >> >> To post message: [email protected] >> List info: http://mail.triembed.org/mailman/listinfo/triembed_triembed.org >> TriEmbed web site: http://TriEmbed.org >> To unsubscribe, click link and send a blank message: >> mailto:[email protected]?subject=unsubscribe >> <[email protected]?subject=unsubscribe> >> >> >> _______________________________________________ >> Triangle, NC Embedded Computing mailing list >> >> To post message: [email protected] >> List info: >> http://mail.triembed.org/mailman/listinfo/triembed_triembed.org >> TriEmbed web site: http://TriEmbed.org >> To unsubscribe, click link and send a blank message: mailto: >> [email protected]?subject=unsubscribe >> >>
_______________________________________________ Triangle, NC Embedded Computing mailing list To post message: [email protected] List info: http://mail.triembed.org/mailman/listinfo/triembed_triembed.org TriEmbed web site: http://TriEmbed.org To unsubscribe, click link and send a blank message: mailto:[email protected]?subject=unsubscribe
