To improve training efficiency, perfect techniques, and train high-quality goalkeepers for the popular sport–football, the use of training equipment such as automatic ball shooting machines is very important. Using the machine to shoot the ball to the desired position in the goal requires a lot of detailed calculations of input parameters and experimentation to check the accuracy of output parameters. Using physics, calculate environmental conditions and forces that affect the ball's trajectory such as Coriolis force, wind force, Magnus effect… etc. From there, theory-based calculated input parameters are obtained: the ball's linear velocity, shot angle and the angular velocity of the machine's two flywheels to shoot the ball to the desired target. To check the accuracy of the above data, build image processing software to redraw the ball's flight trajectory and provide actual output parameters based on the Epipolar geometry method and the 8-point algorithm. With this method, we can find the cause of errors between theoretical calculation and reality, then make changes to input data to improve the accuracy of each shot without using advanced, bulky, and high cost equipments.