Low-cost Real-time Tightly-Coupled GNSS/INS Navigation System Based on Carrier-phase Double- differences for UAV Applications

This paper shows the design and development of a truly low cost attitude determination and navigation system based primarily on non-dedicated mass market GNSS receivers and antennas, aided by MEMS inertial sensors. GNSS/INS integration combines the advantages of absolute satellite-based positioning with the high dynamic performance and data rates of inertial sensors, having become the standard approach in navigation systems for Unmanned Aerial Vehicles (UAV) among other applications. We have used a Tightly-Coupled (TC) algorithm aimed at the integration of GPS and INS devices, designed to exploit the measurements provided by four GNSS receivers and a MEMS IMU. This task is accomplished by forming double differences (DD) of carrier-phase measurements and properly solving the ambiguities on-the-fly. The robustness of the algorithm is enhanced by exploiting unambiguous DD of code-phase observables or un-differenced pseudoranges when carrier phase ambiguity fix is not reliable. In order to limit the unwanted effects of vibration, the INS raw samples (i.e.gyro and accelerometer measurements) are passed through a proper low-pass filter before being used by the fusion algorithm. Furthermore, we have constrained the TC algorithm with the heading information obtained from the DD carrier-phase measurements of four GNSS receivers that are available in the final system. These carrier-phase measurements need to be processed and their ambiguities solved before becoming usable, for which proper integer ambiguity resolution methods are described too. Achieving the realization of the first prototype relies on advances beyond the state-of-the-art in the fields of multi-receiver GNSS-based attitude determination, inertial MEMS/GNSS data fusion and precise calibration of properly selected GNSS antennas. All these advances aim at optimizing the cost, weight, size and power consumption of the overall system to enable its use in one of the most demanding applications as is the control and navigation of small UAV. The feasibility of a real-time implementation has been demonstrated. We have run the designed TC algorithm first onboard a car for validating the implementation by gathering real data in different user configurations (i.e. static and dynamic) and environmental scenarios (opensky, urban canyon, etc.). For performance evaluation of the real-time system the prototype was mounted onboard the target UAV as a sample case of aerial applications.