http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.78.2284&rep=rep1&type=pdf
In this article it say that, because of the dynamics and high complexity of the robot soccer system as well as maneuverability and speed of its robots, the accurate path planning and prediction of moving targets have gained special importance. Several techniques have been proposed for path planning including the use of genetic algorithms and fuzzy logic.
Coupled to the path-planning problem is obstacle avoidance in a dynamic environment. The path plans must be dynamically updated to reflect the changes in the environment, which means that they have to be created in real-time.
Robots have to exhibit basic actions like positioning at a designated location, moving to the ball or blocking an opponent, turning to a desired angle, circling around the ball and shooting the ball into opponent’s goal. Among other factors, the strategy and path planning in a robot soccer game are dependent on the ball position.
The robot's main sensor system is vision, which captures and processes the image at 30 to 60 frames per second. The odd and even scan fields of the interlaced image are processed separately. For this reason a stationary object like the ball is reported at different locations in each frame due to different quantization errors. The errors are compounded by the variation of light intensity from one frame to another. These quantization errors are inherent in the system and have a significant impact on the shooting accuracy even when the ball is stationary. However, the tests carried out on a moving ball are more significant because the interception accuracy suffers when the ball is moving. This is due to the fact that control actions are initiated based on the current ‘static’ state of objects whereas action must be taken based on predicted future positions.
That's all i had read for this week because i need to process all of this information to understand it.
No comments:
Post a Comment