An identification method for individual driver steering behaviour modelled by switched affine systems

Abstract

This paper adresses the issue of modelling and identification of individual driver steering behaviour from a new point of view, incorporating the idea of human motion being built up by an individual and limited repertoire of learned patterns. We introduce a switched affine model structure to explain a measurable motion alphabet in the driving context and show that this leads to a new identification problem that differs from general hybrid identification issues. To solve this problem, we derive a multi-step model output error criterion and propose an algorithm to simultaneously identify switching times and subsystem parameters out of measurable movement data. We show that this algorithm is capable of identifying the true parameters of known systems as well as fitting real movement trajectories even though no a priori information is given about the true system complexity.

Online Identification of Individual Driver Steering Behaviour and Experimental Results

“…The main challenge of the introduced identification problem is the mutual dependency of the parameter estimation quality and the correctly identified switching times. As in case of measurement noise the parameter estimationΘ λi of the ith interval improves with increasing estimation time as long as all measurement data belong to the same subsystem λ i (see [29] for detailed discussion), it is reasonable to push the end of the estimation interval k est as far as possible for each subsystem. On the other hand the estimation interval should not be larger than the true interval [τ i , τ i ] for the active subsystem λ i , because the usage of measured values of the next subsystem λ i+1 will lead to a deteriorating parameter estimation in general.…”

“…In [29] an extended procedure is introduced that is guaranteed to yield the true switching times and subsystem parameters respectively for given ground truth. As already mentioned, this requires the whole measurement data to be available.…”

“…In a concurrent paper we propose an algorithm to identify the subsystem parameters of autonomous switched systems as well as the unknown switching times and we prove that the true values can be calculated correctly under some assumptions on the measurement noise [29]. However, this method is not online-capable because it includes running the algorithm through the measurements forwards and backwards.…”

“…al. we suggest to describe and parameterize dynamic system models that can produce the measurable trajectories as their output [14], [29]. Within this approach the focus is on the system parameters as the inherent quantities of individual human movement.…”

A Driving Behaviour Model of Electrical Wheelchair Users

“…Huge research on joysticks and related interfaces including haptic systems has emerged [ 3 – 7 ], and new control models [ 8 , 9 ] are continuing to develop. The available driver models however suffer lack of individuality [ 10 ], focusing mostly on the common user attributes, and assume that all users respond to particular navigational situations by similar general patterns. Such driver models employ general parameters that barely correspond to measurements obtained from extreme users and hardly take into consideration the contextual nature of human response to stimuli.…”