Robust 3-D Velocity-Constrained Vehicle Positioning System: Immunity to Variances in Smartphone Mounting Angles

The integration of global navigation satellite systems (GNSS) and inertial measurement units (IMUs) forms the cornerstone of advanced smartphone-based navigation solutions. In GNSS-degraded areas such as urban canyons and tunnels, GNSS reliability is severely affected due to signal blockage and reflection, shifting the duty of positioning accuracy to the IMU's performance. However, smartphone IMUs are typically low-quality, leading to rapid positioning accuracy decline. Velocity constraints can enhance IMU performance but face challenges: accurate vehicular velocity acquisition without extra hardware and the need for mounting angle (MA) knowledge, which changes with each installation. To address these issues, we propose an adaptive 3-D velocity-constrained positioning system. This system includes a deep learning-based odometry network for precise velocity estimation and an algorithm that automatically determines the smartphone's MA through the integration of GNSS, IMU, barometer, and map-matching (MM) strategies. Our comprehensive experimental evaluations demonstrate that this innovative approach can substantially improve positioning accuracy by 85% in GNSS-degraded areas.