Enhancing Vehicle Navigation and Safety through Integration of Pre-Recorded Maps with Vehicle Control Unit

Abstract

Powerful navigation and safety systems are vital to ensuring autonomous (DV) or semi-autonomous vehicles continue well in varied environmental conditions. However, in such a world those who depend on that internet connectivity to navigate their car would have problems in areas where network reliability is less than perfect. The purpose of this work is to investigate the feasibility and advantages of combining offline mapping with locally sensed positioning systems for better vehicle navigation and safety. While connected to the internet, vehicles cache offline maps and use local sensor information (such as GPS) from their inertial navigation system or computer vision without needing continuous access. The research outlines a hardware-software architecture with embedded offline map data storage, edge-based road following algorithms and an integration mechanism to advanced driving assistance systems (ADAS). In the evaluation of performance, accuracy assessments with online mapping systems are considered for offline maps and how these can have implications on end driver usability as well impacts on real-world autonomous driving technologies. The results suggest that offline mapping and on-board sensor-based localization would be able to improve vehicle contextual navigation performance in diverse driving scenarios.