Motion State Estimation of Preceding Vehicles With Packet Loss and Unknown Model Parameters

Intelligent connected vehicles can perform better decision making and path planning tasks when they obtain the accurate motion state of preceding vehicles. In this article, an event-triggered estimation scheme considering the effects of packet loss and unknown vehicle inertia parameters is proposed for predicting the motion state of the preceding vehicle in vehicle-to-everything communications. First, an event-triggered communication mechanism is employed instead of the traditional periodic wireless communication rules to enable intermittent transmission of preceding vehicle sensor data to the host vehicle. Further based on the obtained sensor data, an event-triggered cubature Kalman filter (ETCKF) is designed for the host vehicle to estimate the motion state of the preceding vehicle. To further improve the robustness of the algorithm against perturbation of model parameters, a strong tracking algorithm is utilized to optimize the ETCKF to form a strong tracking ETCKF (STETCKF). Finally, the experimental results demonstrate that the estimation accuracy of the STETCKF under the communication rate of 7.08% and model parameters perturbation can be comparable to that of the cubature Kalman filter with 100% communication rate.