[ros-users] camera1394 and swissranger

Radu Bogdan Rusu rusu at willowgarage.com
Fri Apr 2 16:51:16 UTC 2010


No worries. :) I was surprised that my colleagues from TUM have not mentioned that earlier, as the sensors are still 
being used as far as I know.

Do you still want to proceed with modifying the driver to produce both PointCloud and PointCloud2 messages? I've started 
writing tutorials for PCL and basically doing something like a planar segmentation on PointCloud2 will be a single line 
command. You could then provide this as a demo for the SwissRanger driver on its wiki page, because let's face it, ths 
SR is probably going to do give you its best data from planar surfaces rather than anything else :)

Cheers,
Radu.

Patrick Beeson wrote:
> Sorry for the confusion.  I assumed that Mesa's libusb only handled 
> their older devices, and that the new libmesasr was created for the 4000 
> release.
> 
> Radu Bogdan Rusu wrote:
>> Great work Patrick! Thanks.
>>
>> Btw, the swissranger in the TUM repository _was_ already handling both 
>> SR3k and SR4k, as we had both sensors there. If it didn't properly, 
>> someone must have introduced a bug somewhere. I remember Morgan sent 
>> me some patches but due to me moving overseas, I never had time to 
>> check them in.
>>
>> Cheers,
>> Radu.
>>
>> Patrick Beeson wrote:
>>> I have added a new sr4k package to the camera_drivers_experimental 
>>> stack.  This extends the swissranger package in the tum repository to 
>>> handle the newer Mesa Imaging libraries, and supports both SR3000 and 
>>> SR4000 devices.
>>>
>>> I do not have an SR3K device, so I could only test with an SR4K, but 
>>> the API is the same for both (an early calls tells the Mesa libraries 
>>> which device you want to talk to and differences are handles at the 
>>> Mesa library level).
>>>
>>> I have not yet included a standalone viewer, but users can view the 
>>> output in rviz.  The current frame id is the node name (defaults to 
>>> /sr4k).  The user can view Image and PointCloud types.  There are 3 
>>> 2D images, distance, intensity, and confidence.
>>>
>>> Patrick Beeson
>>>
>>

-- 
| Radu Bogdan Rusu | http://rbrusu.com/



More information about the ros-users mailing list