Multi-Antenna Vision-and-Inertial-Aided CDGNSS for Micro Aerial Vehicle Pose Estimation

A system is presented for multi-antenna Carrier Phase Differential GNSS (CDGNSS)-based pose (position and orientation) estimation aided by monocular visual measurements and a smartphone-grade inertial sensor. The system is designed for micro aerial vehicles, but can be applied generally for low-cost, lightweight, high-accuracy, geo-referenced pose estimation. Visual and inertial measurements enable robust operation despite GNSS degradation by constraining uncertainty in the dynamics propagation, improving fixed-integer CDGNSS availability and reliability in areas with limited sky visibility. No prior work has demonstrated an increased CDGNSS integer fixing rate when incorporating visual measurements with smartphone-grade inertial sensing. A central pose estimation filter receives measurements from separate CDGNSS position and attitude estimators, visual feature measurements based on the ROVIO measurement model, and inertial measurements. The filter’s pose estimates are fed back as a prior for CDGNSS integer fixing. A performance analysis under both simulated and real-world GNSS degradation shows that visual measurements greatly increase the availability and accuracy of low-cost inertial-aided CDGNSS pose estimation.