Subject:
|
Re: 2D positioning ???
|
Newsgroups:
|
lugnet.robotics
|
Date:
|
Fri, 10 Dec 1999 14:39:44 GMT
|
Original-From:
|
Tim Gion <tcg3j@SAYNOTOSPAMcs.virginia.edu>
|
Viewed:
|
696 times
|
| |
|
|
On Fri, 10 Dec 1999, Brent Rager wrote:
> With an overhead camera you may find it easiest to use the IR port on the RCX
> to beacon the robots position. With the coordinating PC sending out a poll
> message addressed to individual robots. Most video see IR signals very well
> and a simple frame grabber followed by a scan, similiar to how a light pen
> works (in reverse) could be used for a 2d position. Orientation of the robot
> will be a little more difficult, perhaps another IR source on the far side
> triggered after the inital poll?
In RoboCup, each robot has a ping-pong ball mounted on top. I found one
example of how to find orientation from one of the Cornell teams. They
mounted an additional marker in a fixed offset from the ping-pong ball.
When they did the image processing, they looked for both the ping-pong
ball and the orientation marker to determine the robots' orientation.
> Otherwise you will need to put an easily visible unique pattern on the top of
> each robot, grab a frame, and perform pattern recognition. I don't know if the
> vecoterization software available will work well with the limited resolution of
> a video camera, perhaps if each robot was shaped differently?
If each robot starts in a fixed position, you can match the previous
frame's position of each robot with the closest robot in the current
frame. In this scheme, you could have the same colored markers for each
robot. There could be some correspondence errors with a greedy
association algorithm such as this one. A globally optimal association
algorithm could be used to reduce errors. This is also described in the
CMU paper I mentioned yesterday.
Tim
Tim Gion
tcg3j@cs.virginia.edu (fastest) or gion@cs.virginia.edu
http://www.cs.virginia.edu/~tcg3j
|
|
Message is in Reply To:
 | | Re: 2D positioning ???
|
| (...) task (...) With an overhead camera you may find it easiest to use the IR port on the RCX to beacon the robots position. With the coordinating PC sending out a poll message addressed to individual robots. Most video see IR signals very well and (...) (25 years ago, 10-Dec-99, to lugnet.robotics)
|
8 Messages in This Thread:
  
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
