Abnormal Road Surface Detection Using Wheel Sensor Data

08/20/2021

∙

by   Tamas Dozsa, et al.

∙

0

∙

share

In this manuscript we demonstrate that accurate road abnormality detection based on signals from a 3D force measuring sensor implanted into the tires of a vehicle is possible. We discuss approximating the sensor's output using adaptive Hermite-functions [4] and present an experiment that shows the connection between abnormal road conditions and the level of noise in the residual signal. Finally, we experiment with different classification schemes and conclude that a model-based neural network architecture (VP-NET [7]) outperforms the other candidates in both accuracy and simplicity for surface abnormality detection tasks.