MULTI-GNSS REAL-TIME PRECISE POSITIONING FOR PEDESTRIAN LANE DETECTION

This study investigates the feasibility of using signals from multi-GNSS constellations for lane detection. We created a very precise map of the pedestrian lanes (i.e., athletic tracks inside our sport facility center) using professional, off-the-self, high-end GNSS receiver. These accurate trajectories are considered as ground-truth. A number of real-time kinematic solutions are produced in different scenarios. Both professional and consumer, lower-end, raw-measurement enabled equipment is deployed. Investigations cover the effects on the positioning domain of multi-GNSS signals from such receiver types in combination with different precise positioning algorithms (PPP, RTK and NRTK). The real-time positioning solutions indicate a standard deviation smaller than 0.12 cm (RT-PPP and high-end receiver) and 0.20 cm (NRTK and lower-end receiver). In addition, all real-time solutions are accurate to less than 0.5 m with 68% confidence (1 sigma error). We conclude that in more than 98% of the time the horizontal distance error is smaller than 0.60 m requirement for pedestrian lane detection. This requirement is achieved with professional-grade, high-end multi-GNSS equipment in absolute mode (i.e., PPP algorithm) and navigation-grade, lower-end equipment in differential mode (i.e., RTK/NRTK algorithm).