SIGNAL PROCESSING IMPROVEMENTS TO LOCALIZATION FOR AUTONOMOUS VEHICLES

Precise localization is a fundamental part of vehicular autonomy. Current implementations rely on expensive sensor arrays and favorable conditions to accurately localize a vehicle. As autonomous vehicles progress toward production models, these expensive sensors will be replaced by low-cost alternatives. Additionally, production autonomous vehicles will need to operate reliably in a range of conditions. Localization algorithms will need to accurately perform despite the increased noise due to these factors. This thesis examines mathematical improvements to existing localization techniques in an effort to increase their reliability in adverse conditions. The effectiveness of these improvements is evaluated using numerical simulations as well as testing with real-world data