A Real-Time, Robust Visual-Inertial Navigation System Tightly Coupled With GNSS and Barometer

Visual-inertial navigation systems suffer from accmulated errors and poor robustness. This letter proposes a solution to address these challenges by fusing global navigation satellite system (GNSS) and barometer measurements. Barometer could provides higher measurement precision compared to GNSS in altitude. We have developed a "novel sensor" by creatively integrating measurements from GNSS and barometer. A nonlinear optimization method is used to tightly couple measurements of GNSS-Barometer (GB), visual and inertial information for real-time and robust state estimation. The proposed GB aided visual-inertial navigation system (GB-VINS) is evaluated on extensive large-scale urban driving datasets. GB-VINS demonstrates competitive performance compared with the existing state-of-the-art methods on publicly available datasets.