Multisensor Localization Architecture for High-Accuracy and High-Integrity Land-based Applications

Abstract

Emerging safety-related applications like autonomous vehicles will require high levels of navigation performance in terms of accuracy while still satisfying stringent requirements of integrity, availability and continuity. Achieving sub-meter accuracy in land-based scenarios with GNSS-based solutions can only be achieved with carrier-phase based approaches in combination with additional sensors. So that these solutions can be considered for future certified systems, the safety aspect must be ensured, which is still a challenge in the presence of local GNSS threats (like multipath or NLOS) as well as in multisensor architectures. In this work, we propose a multisensor architecture that uses float Real-Time Kinematic (RTK) GNSS with additional integrity information from an augmentation network. The architecture also considers different layers of protection against local GNSS threats that supports the rigorous design of integrity monitoring algorithms and protection level computation. GNSS is combined with additional sensors like Inertial Measurement Unit (IMU) and a robust relative location of the vehicle is complemented with stereo camera and vision processing. This allows for different possible types of localization modes. The algorithms are validated with real measurements collected with a car during a measurement in Rome, Italy. The results show clearly that our design can achieve high accuracy while ensuring high integrity.