End-to-End Probabilistic Ego-Vehicle Localization Framework

Locating the vehicle in its road is a critical part of any autonomous system and has been subject to different research topics. In most works presented literature, ego-localization split into three parts: Road level-localization consisting on which travels, Lane level localization lane Ego being lateral position ego-lane. For each part, several researches have conducted. However, relationship between parts not taken consideration. Through this work, an end-to-end framework introduced with two main novelties. The first one proposition complete solution that tackles every ego-localization. second lies information-driven approach used. Indeed, we use prior about structure from digital map order reduce space complexity for recognition process. Besides, fusion techniques based Bayesian Network Hidden Markov Model are elaborated leading method is, large extent, robust erroneous sensor data. robustness proposed proven datasets varying scenarios.