Adaptable Vehicle Detection and Speed Estimation for Changeable Urban Traffic With Anisotropic Magnetoresistive Sensors

This paper presents an adaptable roadside vehicle detection and speed estimation system for various traffic conditions on urban roads based on tri-axial anisotropic magnetoresistive sensors and wireless sensor network. The system consists of one master node and two sensor nodes, which are placed along the roadsides and can measure the earth's local magnetic field disturbance caused by passing vehicles. A dynamic threshold detection algorithm is proposed for vehicle detection, especially considering the actual variable traffic condition. The vehicle speed is estimated on the basis of the maximum values and the cross correlation of effective parts extracted from two sensor signals. We have tested the vehicle information at several roads under different traffic conditions. Validation study has revealed a high detection accuracy of 97.92% when using a dynamic threshold compared with 92.3% when using a fixed threshold. And, the average accuracy of speed estimation can reach up to 97.11% on the roads. The proposed algorithm has a significant increase in accuracy, reliability, and practicability compared with the fixed threshold algorithm.