Traffic flow of connected and automated vehicles in Smart Cities: Human-Centric

Abstract

This paper proposes a novel Connected and Automated Vehicle (CAV) model as a scanner for heterogeneous traffic flows, which employs CAV on the road to detect traffic flow characteristics in multiple traffic scenes through various sensors. The model contains the hardware platform and software algorithm of CAV, and the analysis of traffic flow detection and simulation by Flow Project from Mobile Sensing Lab at UC Berkeley and Amazon AWS Machine Learning research grants based on SUMO, where the driving of the car is mainly controlled by Reinforcement Learning (RL). The simulation results showed that the traffic flow scanning, tracking and data recording by CAV are continuous and effective for the wide range and identification confirm function when CAV are in one lane. The effective detection area of CAV is in bow shape in the heterogeneous traffic flow, the occlusion rate is not associated with the lane position of CAV. Therefore, the calculated results should be filtered and optimized to enhance the confidence of heterogeneous traffic data collected. Currently, standards or most practitioners are not aware of this.