Accurate and effective vehicle acceleration prediction is an important function for autonomous vehicles. This paper puts forward an acceleration prediction model based on nonlinear auto regressive models with exogenous inputs (NARX) for on-board implementation. Describing the vehicle’s motion in the horizontal plane requires two accelerations and one angular acceleration. In this paper, the acceleration of two vertical directions and the angular acceleration of the vehicle in the horizontal plane are predicted with the NARX model. This approach effectively captures the timing characteristics of the acceleration of the vehicle. Its nonlinear mapping capability could deal with complex environment during driving. The input of the model is the effective data that is recorded by the vehicle-mounted sensor when the driver is driving. The predictor is trained and tested with vehicles’ real urban driving data. It contains different driving situations. As a result, it improves the generalization ability of this model.
Vehicle Acceleration Prediction Based on Nonlinear Auto Regressive Models with Exogenous Inputs
18th COTA International Conference of Transportation Professionals ; 2018 ; Beijing, China
CICTP 2018 ; 102-112
02.07.2018
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Weather-Weighted Periodic Auto Regressive Models for Sector Demand Prediction
British Library Conference Proceedings | 2009
|