Fault-Free Integrity of Urban Driverless Vehicle Navigation with Multi-Sensor Integration: A Case Study in Downtown Chicago

Abstract

This paper investigates how global navigation satellite systems (GNSSs) and inertial navigation systems (INSs), when appropriately augmented by ranging from local landmarks, can safely navigate vehicles through a real-world urban environment. We begin by considering safety requirements for driverless vehicles under fault-free assumptions and developing measurement models for multi-sensor integrated navigation systems using an extended Kalman filter. The critical elements of urban navigation are then discussed, including individual INS noise parameter specifications, vehicle speed, and the effect of velocity updates. Covariance analyses performed along a 9-km-long urban transect in downtown Chicago show that velocity updates measured by wheel speed sensors, vehicle kinematic constraints, and zero-velocity updates can extend navigation continuity by bridging intermittent GNSS signal availability. However, position reference updates at intervals between 15 and 35 m, based on light detection and ranging data from local landmarks in our case, are needed to achieve full navigation availability through the transect.