View Full Version : Camera-to-Driver Station latency
01-10-2012, 12:13 PM
Does anyone have an accurate characterization of the latency we can expect in transmitting an image (say, 320x240 with compression set to 30) from the Axis camera to the Driver Station? More than a second? Less than 1/2 a second? Near instantaneous?
Our team, like many others, is weighing our options when it comes to vision processing and, while laptop-side processing is attractive and latency is low under our controlled conditions, we are concerned about what kind of latency to expect under competition circumstances.
01-10-2012, 12:47 PM
Unfortunately, I don't have a number for this.
Are you interested in FPS or latency? I'd imagine that with this system, they are mismatched a bit.
01-10-2012, 02:12 PM
Well, both are nice. But latency is more critical to the way we are thinking currently.
01-10-2012, 02:13 PM
I don't have a recent measure, but I have measured it in the past by enabling an LED (assumed to be an instantaneous stimulus), and doing very cheap pixel intensity test on the controller, laptop, or wherever you would like to determine the latency.
The camera has its own internal clock and that will use that clock to start and stop exposures -- no more often than every 33ms. We have no good way to synchronize or control that clock, so the code will make numerous measurements possibly with intentional random jitter, and look at it statistically to find the distribution of times. This will give you a range of expected lag for a particular setup.
I'll see if I can redo the measurements to the cRIO and to the dashboard and post the distribution graphs.
The FPS numbers are far easier to measure and are independent of the lag.
vBulletin® v3.6.4, Copyright ©2000-2013, Jelsoft Enterprises Ltd.