Autonomous driving is developing from an insurmountable aspiration into a more realistic vehicle's feature radically transforming the transportation system. Implementing a self-localization system with an accurate performance is a fundamental and substantial challenge for accomplishing this objective. Global navigation satellite systems are frequently implemented to register the absolute position and orientation of the vehicle. However, in the absence of satellite signals, an alternative is required to ensure the unimpaired functionality of the self-localization system. A widely applied method with a straightforward implementation is dead reckoning as a self-localization technique based on mechanical motion sensors. A major disadvantage of such a system is the self-localization inaccuracy due to the sensing system errors propagating over time. Access to an environment map allows the application of registration techniques using environment perception through appropriate sensing systems. In this book, the realization of the map registration algorithms is performed with radars, as they perceive the environment regardless of weather and light conditions. New radar-based self-localization solutions are conceptualized and implemented for GNSS-restricted areas utilizing registration of environment models and sequential Monte-Carlo method. A map registration algorithm is proposed utilizing radar observations and OpenStreetMap initialized with dead reckoning. Moreover, a particle filtering approach is implemented to realize absolute self-localization with divergence avoidance strategies. As the preeminent contribution of this book, the map registration of radar measurements is transferred from Cartesian coordinates to Range-Doppler coordinates offering significant advantages.weiterlesen