A COMPARATIVE ANALYSIS OF LOW COST LOCALIZATION METHODS FOR A MARIO KART GAME APPLICATION
In vehicle navigation systems, it is important to have accurate estimates of vehicle positioning. The process of determining positioning is called localization. There are two methods for localization, absolute positioning and relative positioning [1]. An absolute positioning system, such as Global...
| Main Author: | |
|---|---|
| Format: | Others |
| Published: |
DigitalCommons@CalPoly
2012
|
| Online Access: | https://digitalcommons.calpoly.edu/theses/841 https://digitalcommons.calpoly.edu/cgi/viewcontent.cgi?article=1887&context=theses |
| Summary: | In vehicle navigation systems, it is important to have accurate estimates of vehicle positioning. The process of determining positioning is called localization. There are two methods for localization, absolute positioning and relative positioning [1]. An absolute positioning system, such as Global Positioning System (GPS), determines the localization by using external distance and heading measures. Relative localization determines the position of an object based on its previously known location and through sensors that determine speed and heading. Examples of relative positioning sensors include wheel encoders, inertial sensors and accelerometers.
The objective of this thesis is to analyze different methods to perform localization and navigation on low speed, low cost go-karts for Augmented Reality Mario Kart game application. Several different low cost localization methods (both absolute and relative positioning methods) were tested and compared in order to improve positioning information of the go-karts.
It was found that the popular solution for localization, absolution positioning using GPS, alone provided insufficient accuracy needed for this game system. Dead-Reckoning (DR), a relative positioning technique, also suffered from inaccuracies due to accumulation of measurement error when used by itself. The optimum low cost solution was achieved by using GPS and DR in conjunction with beacons to deliver the best accuracy. The improvement of this solution with four beacons over the standalone GPS solution was 3.620 meters in Root Mean Square Error (RMSE), the error criterion used in this project. |
|---|