A Model-based Tightly Coupled Architecture for Low-Cost Unmanned Aerial Vehicles for Real-Time Applications

This paper investigates the navigation performance of a vehicle dynamic model-based (VDM-based) tightly coupled architecture for a fixed-wing Unmanned Aerial Vehicle (UAV) during a global navigation satellite system (GNSS) outage for real-time applications. Unlike an Inertial Navigation System (INS) which uses inertial sensor measurements to propagate the navigation solution, the VDM uses control inputs from either the autopilot system or direct pilot commands to propagate the navigation states. The proposed architecture is tested using both raw GNSS observables (Pseudorange and Doppler frequency) and Micro-Electro-Mechanical Systems-grade (MEMS) Inertial Measurement Unit (IMU) measurements fused using an extended Kalman filter (EKF) to aid the navigation solution. Other than the navigation states, the state vector also includes IMU errors, wind velocity, VDM parameters, and receiver clock bias and drift. Simulation results revealed significant performance improvements with a decreasing number of satellites in view during 140 seconds of a GNSS outage. With two satellites visible during the GNSS outage, the position error improved by one order of magnitude as opposed to a tightly coupled INS/GNSS scheme. Real flight tests on a small fixed-wing UAV show the benefits of the approach with position error being an order of magnitude better as opposed to a tightly coupled INS/GNSS scheme with two satellites in view during 100 seconds of a GNSS outage.