Legacy Forum: Preserving Nearly 20 Years of Community History - A Time Capsule of Discussions, Memories, and Shared Experiences.

OpenCV in humanoid

Anything that doesn't fit our other forums goes here.
11 postsPage 1 of 1
11 postsPage 1 of 1

OpenCV in humanoid

Post by robaldo » Tue Dec 28, 2010 9:15 pm

Post by robaldo
Tue Dec 28, 2010 9:15 pm

hi all
have you ever used OpenCV for image processing in humanoid robots?
i want to know how OpenCV help robotic and robocup (Humanoid)!
how to use it!?!
is using OpenCV better than learning image processing basics?
why? what are benefits of learning OpenCv?
thanks
hi all
have you ever used OpenCV for image processing in humanoid robots?
i want to know how OpenCV help robotic and robocup (Humanoid)!
how to use it!?!
is using OpenCV better than learning image processing basics?
why? what are benefits of learning OpenCv?
thanks
robaldo
Savvy Roboteer
Savvy Roboteer
Posts: 53
Joined: Fri Dec 03, 2010 4:49 pm

Post by SK » Wed Dec 29, 2010 10:45 am

Post by SK
Wed Dec 29, 2010 10:45 am

Most RoboCup teams (in the humanoid and standard platform league, that is) use largely custom image processing approaches, as there is very limited processing power available on the robots and it thus makes sense to use all available knowledge and possible optimizations for the given scenario.
In both leagues, elements of the playing field and robots are currently color-coded and this fact is leveraged in image processing systems. Both B-Human and the Dribblers (champions 2009 and 2010 in their respective leagues) use a so called "scanline" approach, where the image gets sampled using scanlines that are more coarse than the image resolution. For all sample points, the pixel gets assigned a color class using a look-up table and based on this, with further processing, objects can be detected.
Of course, the playing field will not stay color coded forever and the rules are moving a small step towards a real soccer scenario every year, so all that will change over the years and with more processing power available in smaller packages we'll likely also see more use of OpenCV in RoboCup humanoids.
All that of course does not mean that OpenCV is "bad", it only means that with the severe constraints of the RoboCup humanoid scenario, custom solutions are currently more successful.
Most RoboCup teams (in the humanoid and standard platform league, that is) use largely custom image processing approaches, as there is very limited processing power available on the robots and it thus makes sense to use all available knowledge and possible optimizations for the given scenario.
In both leagues, elements of the playing field and robots are currently color-coded and this fact is leveraged in image processing systems. Both B-Human and the Dribblers (champions 2009 and 2010 in their respective leagues) use a so called "scanline" approach, where the image gets sampled using scanlines that are more coarse than the image resolution. For all sample points, the pixel gets assigned a color class using a look-up table and based on this, with further processing, objects can be detected.
Of course, the playing field will not stay color coded forever and the rules are moving a small step towards a real soccer scenario every year, so all that will change over the years and with more processing power available in smaller packages we'll likely also see more use of OpenCV in RoboCup humanoids.
All that of course does not mean that OpenCV is "bad", it only means that with the severe constraints of the RoboCup humanoid scenario, custom solutions are currently more successful.
SK
Savvy Roboteer
Savvy Roboteer
User avatar
Posts: 105
Joined: Mon Jun 26, 2006 1:00 am
Location: near Frankfurt/Main, Germany

Post by robaldo » Fri Dec 31, 2010 7:09 am

Post by robaldo
Fri Dec 31, 2010 7:09 am

SK wrote:In both leagues, elements of the playing field and robots are currently color-coded and this fact is leveraged in image processing systems. Both B-Human and the Dribblers (champions 2009 and 2010 in their respective leagues) use a so called "scanline" approach, where the image gets sampled using scanlines that are more coarse than the image resolution. For all sample points, the pixel gets assigned a color class using a look-up table and based on this, with further processing, objects can be detected.


thanks
could you please explain more about scanlines?
i want to know weather i can do image processing with limited hardware or not!?
thanks in advance
SK wrote:In both leagues, elements of the playing field and robots are currently color-coded and this fact is leveraged in image processing systems. Both B-Human and the Dribblers (champions 2009 and 2010 in their respective leagues) use a so called "scanline" approach, where the image gets sampled using scanlines that are more coarse than the image resolution. For all sample points, the pixel gets assigned a color class using a look-up table and based on this, with further processing, objects can be detected.


thanks
could you please explain more about scanlines?
i want to know weather i can do image processing with limited hardware or not!?
thanks in advance
robaldo
Savvy Roboteer
Savvy Roboteer
Posts: 53
Joined: Fri Dec 03, 2010 4:49 pm

Post by SK » Mon Jan 03, 2011 9:19 am

Post by SK
Mon Jan 03, 2011 9:19 am

The German Team Report 2005 has a pretty good explanation of how it was done for the Aibos: http://www.germanteam.org/
A PDF as well as the source code can be downloaded there.

A similar release for 2010 is available from team B-Human in the Standard Platform League:
http://www.b-human.de/en/publications/
The German Team Report 2005 has a pretty good explanation of how it was done for the Aibos: http://www.germanteam.org/
A PDF as well as the source code can be downloaded there.

A similar release for 2010 is available from team B-Human in the Standard Platform League:
http://www.b-human.de/en/publications/
SK
Savvy Roboteer
Savvy Roboteer
User avatar
Posts: 105
Joined: Mon Jun 26, 2006 1:00 am
Location: near Frankfurt/Main, Germany

Post by SK » Wed Feb 23, 2011 8:52 am

Post by SK
Wed Feb 23, 2011 8:52 am

Small update: We uploaded a short "first person view" video of Bruno kicking a goal. It also shows the detections by the image processing system (even if probably a bit hard to see in the video, as it looks rather hectic to a human :) ).

phpBB [media]
Small update: We uploaded a short "first person view" video of Bruno kicking a goal. It also shows the detections by the image processing system (even if probably a bit hard to see in the video, as it looks rather hectic to a human :) ).

phpBB [media]
SK
Savvy Roboteer
Savvy Roboteer
User avatar
Posts: 105
Joined: Mon Jun 26, 2006 1:00 am
Location: near Frankfurt/Main, Germany

Post by robaldo » Wed Feb 23, 2011 2:11 pm

Post by robaldo
Wed Feb 23, 2011 2:11 pm

is that the real camera fps of Bruno? is it 30 fps in 640x480 resolution?
also i want to know the circle and line detections are performed by OpenCV or not!?
thanks
is that the real camera fps of Bruno? is it 30 fps in 640x480 resolution?
also i want to know the circle and line detections are performed by OpenCV or not!?
thanks
robaldo
Savvy Roboteer
Savvy Roboteer
Posts: 53
Joined: Fri Dec 03, 2010 4:49 pm

Post by SK » Wed Feb 23, 2011 3:35 pm

Post by SK
Wed Feb 23, 2011 3:35 pm

Yes, we're using a SPC-1300 in 640x480 YUV422 mode at 30Hz. As I said, we use a custom color-based grid/scanline based approach, so no OpenCV is used here.
Yes, we're using a SPC-1300 in 640x480 YUV422 mode at 30Hz. As I said, we use a custom color-based grid/scanline based approach, so no OpenCV is used here.
SK
Savvy Roboteer
Savvy Roboteer
User avatar
Posts: 105
Joined: Mon Jun 26, 2006 1:00 am
Location: near Frankfurt/Main, Germany

Post by robaldo » Wed Feb 23, 2011 5:50 pm

Post by robaldo
Wed Feb 23, 2011 5:50 pm

SK wrote:Yes, we're using a SPC-1300 in 640x480 YUV422 mode at 30Hz. As I said, we use a custom color-based grid/scanline based approach, so no OpenCV is used here.


thanks
could you please tell me what is your main processor?
i mean which board are u using for your robots?
SK wrote:Yes, we're using a SPC-1300 in 640x480 YUV422 mode at 30Hz. As I said, we use a custom color-based grid/scanline based approach, so no OpenCV is used here.


thanks
could you please tell me what is your main processor?
i mean which board are u using for your robots?
robaldo
Savvy Roboteer
Savvy Roboteer
Posts: 53
Joined: Fri Dec 03, 2010 4:49 pm

Post by robaldo » Wed Feb 23, 2011 6:06 pm

Post by robaldo
Wed Feb 23, 2011 6:06 pm

SK wrote:The German Team Report 2005 has a pretty good explanation of how it was done for the Aibos: http://www.germanteam.org/
A PDF as well as the source code can be downloaded there.

A similar release for 2010 is available from team B-Human in the Standard Platform League:
http://www.b-human.de/en/publications/


also i can't find the pdf you mentioned!
please link me directly to it!
i'm eager to learn what the scanline is.
SK wrote:The German Team Report 2005 has a pretty good explanation of how it was done for the Aibos: http://www.germanteam.org/
A PDF as well as the source code can be downloaded there.

A similar release for 2010 is available from team B-Human in the Standard Platform League:
http://www.b-human.de/en/publications/


also i can't find the pdf you mentioned!
please link me directly to it!
i'm eager to learn what the scanline is.
robaldo
Savvy Roboteer
Savvy Roboteer
Posts: 53
Joined: Fri Dec 03, 2010 4:49 pm

Post by SK » Thu Feb 24, 2011 8:07 am

Post by SK
Thu Feb 24, 2011 8:07 am

robaldo wrote:also i can't find the pdf you mentioned!
please link me directly to it!
i'm eager to learn what the scanline is.

http://www.germanteam.org/GT2005.pdf and
http://www.b-human.de/file_download/33/ ... elease.pdf
It really wasn't hard to find though ;)

We're using a FitPC2 (Atom 1.6 GHz) for high level cognition stuff.

Just yesterday the teams qualified for the RoboCup Humanoid League were announced, so you can find a lot of information here:
http://www.tzi.de/humanoid/bin/view/Website/Teams2011
(Our TDP also has a nice overview about our robot hard- and software)
robaldo wrote:also i can't find the pdf you mentioned!
please link me directly to it!
i'm eager to learn what the scanline is.

http://www.germanteam.org/GT2005.pdf and
http://www.b-human.de/file_download/33/ ... elease.pdf
It really wasn't hard to find though ;)

We're using a FitPC2 (Atom 1.6 GHz) for high level cognition stuff.

Just yesterday the teams qualified for the RoboCup Humanoid League were announced, so you can find a lot of information here:
http://www.tzi.de/humanoid/bin/view/Website/Teams2011
(Our TDP also has a nice overview about our robot hard- and software)
SK
Savvy Roboteer
Savvy Roboteer
User avatar
Posts: 105
Joined: Mon Jun 26, 2006 1:00 am
Location: near Frankfurt/Main, Germany

Re: OpenCV in humanoid

Post by arronlee » Sun Sep 29, 2013 11:08 am

Post by arronlee
Sun Sep 29, 2013 11:08 am

robaldo wrote:hi all
have you ever used OpenCV for
image processing in humanoid robots?
i want to know how OpenCV help robotic and robocup (Humanoid)!
how to use it!?!
is using OpenCV better than learning
image processing basics?
why? what are benefits of learning OpenCv?
thanks

Hi, I am a newbie here. I wonder what's the latest version. I am almost a green hand on this field. Thanks in advance.



Bset regards,
Arron
robaldo wrote:hi all
have you ever used OpenCV for
image processing in humanoid robots?
i want to know how OpenCV help robotic and robocup (Humanoid)!
how to use it!?!
is using OpenCV better than learning
image processing basics?
why? what are benefits of learning OpenCv?
thanks

Hi, I am a newbie here. I wonder what's the latest version. I am almost a green hand on this field. Thanks in advance.



Bset regards,
Arron
arronlee
Newbie
Newbie
Posts: 6
Joined: Fri Jul 19, 2013 3:00 am


11 postsPage 1 of 1
11 postsPage 1 of 1