A longitudinal driver model for long horizon speed prediction in powertrain applications

Two major emerging fields of research in automotive engineering are autonomous driving and electromobility. The predictive or intelligent longitudinal control within the former and consumption forecasts for the latter are dependent on lookahead data provided by cloud based services as real time road and traffic data. Furthermore, these applications can be improved by customization to the driver. This paper proposes a simple, yet accurate parametric model for longitudinal driving characteristics which is designed for use in powertrain applications that could be a predictive, intelligent cruise controller or a personalized consumption forecast. A methodology for offline identification of the model parameters is presented that can be easily transferred to online implementation. The model is validated against measurement data and meaningful metrics for assessing its performance are introduced. It is shown that predicted speed yields good resemblance to measurements.