Real-Time Traffic Speed Estimation with Adaptive Cruise Control Vehicles and Manual Vehicles in a Mixed Environment

Abstract

With the development of adaptive cruise control (ACC) vehicles, the traffic on freeways will be a mixture of ACC vehicles and traditional manual vehicles. It is a new challenge for freeway traffic state estimation. This paper proposes a neural network-based method to estimate the traffic speed under the mixed traffic. Three different neural network models are investigated using the simulation data, which is programmed by MATLAB. The results indicate that the neural network model with three input neurons (average speed of ACC vehicles, speeds from the microwave detector, and percentage of ACC vehicles) has a significant performance. Besides the fusion of measurements from the microwave detectors and ACC vehicles could improve the overall estimation accuracy, especially, the penetrate rate of ACC vehicles which is below 35%.