A study of appropriate model complexity for estimation of car-following behavior

Paper i proceeding, 2015

This paper investigates model complexity, or equivalently, necessary model structure, for describing car-following behavior. The recently suggested PrARX modeling framework is systematically investigated and compared with simpler model structures containing less tuning parameters to see to what extent complex models can be motivated. The results indicate that for long prediction horizons there is no gain having a complex model structure. However, one-step ahead prediction of complex models can be useful for classification of whether the driver should be braking or not.