Phondeenana, P., Thitipatanapong, R., Klongnaivai, S., Noomwongs, N. et al., "Driver Behavior Detection based On PPP-GNSS Technology," SAE Technical Paper 2014-01-2006, 2014, doi:10.4271/2014-01-2006.
Driver behavior is one of the most important factors in safe mobility. In general, various driver maneuvers can be determined from acceleration of the vehicle. Physically, the acceleration and brake can be detected with longitudinal acceleration while turning and lane change can be detected with lateral acceleration. Normally, IMU (inertia measurement unit) has been designated to get these data. However, the IMU is not convenience to install in the vehicles especially as aftermarket parts. Nowadays, navigation system technologies have been much improved, both on availability and accuracy with combination of multiple navigation satellite systems. Normally, it's called Multi-GNSS (multiple global navigation satellite system). In particular, the satellite navigation systems available in this work are GPS, GLONASS, and QZSS. With decimeter precision and the update rate scale up to 10-Hz, the GNSS can be a viable alternative for driver behavior detection. In previous studies, it was found that dangerous behaviors such as aggressive lane change required decimeter accuracy from GNSS. Consequently, in this article, the PPP (Precise Point Positioning) technique was applied to develop a behavior detection system. Vehicle maneuvers such as cornering and lane change detected by IMU and Multi-GNSS were collected and compared. The JAVAD Delta TRE_G3TH experimental Multi-GNSS receiver installed on a utility vehicle was used in this work. The navigation data were post-processed with the open source RTKLIB software. On the other hand, the reference vehicle dynamic measurement system was VBOX 3LSi with integrated IMU. A simple curvature model was proposed and applied to estimate the lateral acceleration. In conclusion, experimental results show that driver behavior detection is feasible with the Multi-GNSS PPP. The results revealed that with PPP solution the estimated lateral acceleration could be manipulated within 10% deviation.