Football Robot Games are regularly held by the Ministry of Research, Technology, and Higher Education from year to year through Indonesian Robot Contest (KRI). Several divisions of the robot competition that were contested, one of which was the Indonesian Football Robot Contest (KRSBI) on wheels. Each team that competes must prepare three robots consisting of one goalkeeping robot and two robots as attackers. All robots work automatically to be able to find, dribble, grab, kick and even block the ball. The goal is to win the world cup match.
One of the skills every robot must have is to recognize which ball is. Like humans, the camera on the robot serves as the eye and the mini PC as the brain. The quality of the camera and mini PC also determines the robot’s performance. The human way of thinking in recognizing the ball is taught to robots through program language. This method is known as “Artificial Intelligence” or artificial intelligence.
To recognize the ball it takes a training stage. At the training stage, the robot is introduced to various ball positions, from the ball at close range, long-distance, the ball that is partially covered even if other objects have the same color as the ball.
The next stage is to test how effective the exercise is in recognizing the ball by providing an image of the ball that has never been trained on a robot before. If the test results show high accuracy, the robot is declared ready to recognize the ball. Otherwise, if not, then there is further evaluation to improve accuracy.
In making a robot to recognize a ball, the artificial intelligence embedded in the robot must work with high accuracy.
Author: Winarno, MT
Link: http://journal.uad.ac.id/index.php/TELKOMNIKA/article/view/15009
(Object detection for KRSBI robot soccer using PeleeNet on omnidirectional camera)
