This paper describes a model-based pothole detection algorithm that exploits a multi-phase dynamic model. The responses of hitting potholes are empirically broken down into three phases governed by three simpler dynamic system sub-models. Each sub-model is based on a rigid-ring tire and quarter-car suspension model. The model is validated by comparing simulation results over various scenarios with FTire, a commercial simulation software for tire-road interaction. Based on the developed model, a pothole detection algorithm with Unscented Kalman Filter (UKF) and Bayesian estimation is developed and demonstrated.
展开▼
