next up previous
Next: Data Association Up: Real-Time Perception for Multiple Previous: Real-Time Perception for Multiple

Color-based Detection

The RoboCup rules specify well defined colors for different objects in the field and these are used as the major cue for object detection. Our vision-processing is therefore color based. Teammates and opponents are identified by blue and yellow circles. We add an additional pink patch for each of our robots to determine teammate orientation. The ball is an orange golf ball (see Figure 1).

Noise is inherent in all vision systems. False detections in the current system are often of a magnitude of 100 spurious detections per frame. The system eliminates false detections via two different methods. First, color patches of size not matching the ones on the robots are discarded. This technique filters off most ``salt and pepper'' noise. Second, by using a minimum-distance data association mechanism, all false detections are eliminated.

Peter Stone
Sun Dec 7 06:55:46 EST 1997