We investigate the shot-aiming problem in Pinball and design a model predictive controller to take shots toward targets with minimal trajectory error using a k-Nearest Neighbor regression-based flip function. A flip function maps the ball's state to its launch trajectory parameters. The k-Nearest Neighbor regression algorithm is applied to reduce the number of calibration shots required to estimate the flip function accurately. The controller is developed for a simulator built with Visual Pinball. The test results from this controller are compared to a controller built with a polynomial flip function. The experiments showed that the controller could accurately aim shots toward targets with minor trajectory errors after a few initialization shots.
Real-Time Model Predictive Control for Shot Aiming in Pinball Using k-NN Regression
28.08.2023
1725927 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Predictive Algorithm For Aiming An Antenna
NTRS | 1993
|Real-Time Estimation Of Aiming Error Of Spinning Antenna
NTRS | 1992
|Tema Archiv | 1987
|Model predictive flight control using adaptive support vector regression
Tema Archiv | 2010
|