Development of ball direction prediction system for wheeled soccer robot goalkeeper using trigonometry technique and neural network method

Abstract

In this research, trigonometry technique and Neural Network method were implemented to predict the ball direction for wheeled soccer robot goalkeeper. The performance of goalkeeper robot in Wheeled Soccer Robot Contest is very important. The crucial problem with goalkeeper robot is the delay in ball detection by the camera because the results of the captured images are always slower than the pictures that have been captured. This causes the robot response to block the opponent's kicked ball being late. Trigonometry technique is one technique that can be used to predict the direction of the ball based on trigonometric mathematical formulas. The used input variables are the location of the last ball position (x–last ball and y-last ball) and the location of the current ball position (x-current ball and y-current ball). In this system a Neural Network method is also implemented to estimate the ball distance from goalkeeper robot. The result shows the goalkeeper robot successfully predicts the ball direction very well and it can estimate the ball distance with 7.06 cm error accuracy. By implementing this method can optimize the performance of the goalkeeper robot in saving the goal.