Evaluating INS/GNSS Availability for Self-Driving Cars in Urban Environments

We evaluate fault-free integrity of tightly coupled INS/GNSS-based navigation for a self-driving car in a dense urban environment. Specifically, we estimate how long and under what local conditions a vehicle’s navigation protection level does not exceed an alert limit in a real urban environment—in our case, 3D-mapped downtown Chicago. We use a practical integrated navigation system, consisting of GNSS, INS, wheel speed sensors, and vehicle dynamic constraints. The goal is to determine at what locations augmentation from external ranging sources is needed to maintain continuous navigation with fault-free integrity. The results show that the vehicle velocity passing through GPS-denied areas strongly affects INS/GPS performance, and that wheel speed sensors and vehicle dynamic constraints can provide some relief.