Cooperative Positioning Enhancement for HDVs and CAVs Coexisting Environment Using Deep Neural Networks

Ao Zhang, Xuting Duan, Jianshan Zhou, Haiying Xia, Long Zhang, He Liu, Wenjuan E

Abstract: Accurate vehicle positioning is a key technology affecting traffic safety and travel efficiency. High precision positioning technology combined with the internet of vehicles (IoV) can improve the positioning accuracy of human-driving vehicles (HDVs), which is well suited for practical application requirements and resources saving. In this paper, a positioning error prediction model based on deep neural network (DNN) and positioning information sharing methods are proposed for traffic scenarios where connected and autonomous vehicles (CAVs) and HDVs with different positioning capabilities coexist. The CAVs with high precision positioning capability is utilized to share positioning information for HDVs to enhance the cooperative positioning accuracy of vehicles with different positioning capabilities. Experimental results show the accuracy and timeliness of our proposal for enhancing vehicle positioning accuracy and sharing vehicle positioning information.