State Estimation Predictor with Stochastic time-delay Based Framework for Teleoperated Vehicle

The recent automotive industry has consistently focused on autonomous driving technologies for their potential chances to free human drivers from tedious maneuvering labor. Although commercialized full-driving automation vehicles are thought to be capable of replacing human drivers, many disengagement reports of autonomous driving are doubtful about its safety-integrity levels to fulfill the acceptable minimum risk. As a countermeasure and backup plan for the disengagement of autonomous driving, recent approaches have focused on the teleoperation method in which humans and vehicles collaborate. However, the performance of the teleoperation method largely relies on the communication quality between the control center and the vehicle. Due to this fact, the communication delay has to be preliminarily investigated and profiled to be used as a boundary to classify as a communication outlier. Therefore, we developed a real-time communication delay classification scheme using a state estimation predictor with a stochastic time-delay-based Framework. The proposed framework was tested with an actual human-vehicle interface using a lte/5G network with a global navigation satellite (GNSS) system. The communication delay was classified according to the level of the potential risk considering the magnitude of the communication delay. The result showed that the total amount of the communication delay was compensated by considering the communication time delay.