Thanks Bill.  This slipped through my email filters.  Let me know 
if/when you figure this out.  I'd be happy to test and apply a working 
patch.

On 05/06/2010 03:25 PM, Bill Morris wrote:
> On Thu, 2010-05-06 at 13:34 +0200, Ulrich Friedrich Klank wrote:
>> Hi Bill,
>>
>> I made the experience that image transport over USB usually introduces a
>> relatively high lag. Given your camera is on a bus with lower latency
>> (like firewire), you might just have an constant temporal offset.
>
> The camera is also USB, so that's why the lag doesn't look like a bus or
> cpu issue.
>
>> In dev_sr.cpp at line 196 the assumption is made, the the image is
>> acquired synchronously, meaning, that function SR_Acquire trigger the
>> acquisition of an image on the swissranger and transport this image back
>> to the pc, which might not be correct. More probably SR_Acquire just
>> copies the last image on the usb communication stack of the driver to
>> the user memory, or waits for the next arriving image, if there is
>> nothing in queue.
>
> Yes, I believe the swissranger_camera driver is only setup for
> continuous mode operation. Our communications with Mesa so far seems to
> indicate that there is some queuing happening with the transmission.
>
>> You most probably can fix your problem by adding a constant offset to
>> timestamp at this point in the file depending on the overall framerate
>> and the transportation time over the usb-bus.
>>
>> A Calibration of this offset would require most probably a lot of
>> accurate hardware to produce a unique signal in the the swissranger for
>> a precisesly known time. Sorry to not be able to tell anything about
>> this delay, since we currently use the swissranger only under a static
>> world assumption.
>
> I'm just trying to figure out where the delay is coming from at this
> point. Mesa Imaging seems pretty responsive so far and hopefully can
> help sort this out.
>
>> Please keep us informed if you find out about this latency.
>> Good luck,
>> Ulrich Klank
>
> I'll submit patches when we fix the problem.
>
> Also, here is a video showing the issue we are having.
> http://www.youtube.com/watch?v=_37S3q2Z-qk
>
> -Bill
>
> _______________________________________________
> ros-users mailing list
> ros-users@code.ros.org
> https://code.ros.org/mailman/listinfo/ros-users