Speaker
Description
With the rapid development of automated driving, real-time vehicle speed estimation can assist in monitoring road traffic condition and predicting change in traffic volume. Direct transmission of detected traffic information to vehicles in real-time is referred to as Vehicle-to-Infrastructure (V2I) communication, which is essential in automated driving. Magnetic sensor technology, due to its low cost, compact size, strong anti-interference capability, and absence of privacy concerns, is being utilized in new-generation vehicle detection [1]. The prevailing method for speed estimation mainly relies on the time difference between data from at least two sets of sensors as a reference. The simplest instantaneous speed calculation model can be summarized as $v=l/\Delta t$, where $l$ is the distance between sensors, and the $\Delta t$ is time difference of the maximums of two waveforms.
In this study, we utilized 2-axis Magneto-Impedance (MI) sensors for estimate vehicle speed [2]. We placed MI sensors on the side of the road to reduce the interference of vehicle body. An analysis of issues causing a decrease in accuracy has been conducted as shown in Fig.1. Due to interference caused by the structure of the vehicle, the maximum value of the waveform may shift, leading to significant errors in speed calculation. We proposed bandpass filter (0.1-50 Hz) and moving average ($n$=1500) to filter the noise. The dashed line represents the waveform smoothed by a moving average with $n$=1500, where it can be observed that deviations caused by noise are eliminated, enabling more precise calculation of instantaneous speed. This method achieved 97 $\%$ average accuracy in speed estimation ($\pm$ 3 km/h) with sensor spacing of 1.6 m. Additionally, we proposed a speed estimation method based on frequency domain features, validating the feasibility of speed estimation with a single sensor.
Fig. 1 Using the maximum value differential method for speed estimation.
References
[1] S. -L. Jeng, W. -H. Chieng and H. -P. Lu, "Estimating Speed Using a Side-Looking Single-Radar Vehicle Detector," in IEEE Transactions on Intelligent Transportation Systems, vol. 15, no. 2, pp. 607-614, April 2014, doi: 10.1109/TITS.2013.2283528.
[2] R. Yao and T. Uchiyama, "Analysis of Magnetic Signatures for Vehicle Detection Using Dual-Axis Magneto-Impedance Sensors," in IEEE Sensors Journal, vol. 24, no. 6, pp. 8721-8730, March 15, 2024, doi: 10.1109/JSEN.2024.3357852.
