Estimating vehicle speed from road surface vibration using exponential regression

Vehicles' motion on roads causes road surface vibration. Road surface vibration induced by passing vehicles has been investigated by some researchers in attempt to obtain traffic data such as volume, speed, occupancy and vehicle types. This type of non-deterministic study of vibration signals employs parametric as well as non-parametric methods to explore important properties of the vibration signals to estimate traffic data. In this paper, we present non-parametric estimation of vehicle speed from road surface vibration signals induced by passing vehicles. A data logger which comprises of an ADXL355 accelerometer is developed to measure and store the axis acceleration values induced by surface vibration caused by passing vehicles. Root-mean-square acceleration (g-rms) values from the x-axis and y-axis signals of the accelerometer are calculated. By applying exponential regression, the relationship of g-rms of the signals versus the vehicle speed is obtained. The results show acceptable exponential regression with R² exceeding 0.9.