L. Gianone

Approximate H/sub /spl infin// identification using partial sum operators in a disc algebra basis

H/sub /spl infin// system identification using a basis in the disc algebra is presented. The approximate model is represented by a partial sum with respect to this basis. The identification problem is to estimate the expansion coefficients of this partial sum. Since the constructed basis functions cannot be represented analytically, they are approximated in order to arrive at a model in a suitable form. An algorithm is presented which calculates the model parameters from the frequency domain data set.

Automatic Detection of the Lane Departure of Vehicles

The paper reports on the development of an experimental warning system for the visual detection of the unintentional lane departure of road vehicles. The hardware system assembled from available low-cost parts and a fundamental adaptive algorithm are described. The system has been placed on a test vehicle and tested on motorways in real situations by daylight and by night. The results are illustrated with pictures selected from the records saved during the experiments.

H/sub /spl infin// vibration control of active suspension for automotive application

This paper deals with simultaneous control of acceleration transmission, wheel load fluctuation, suspension deflection and sprung mass displacement. The control theory is based on the robust H/sub /spl infin// control theory with the consideration of the modeling errors lumped into an unstructured uncertainty acting on the unperturbed plant. The design method is new and powerful for electronic car suspensions control systems, as it presents a new approach to the designers for the computation of a controller maintaining robust stability while guaranteeing the performance in a wide frequency range, even in the case of the controller is applied to the actual plant. Additionally, the paper emphasizes that the assumptions in the modeling step may lead to unrealistic control designs and, as an example, the effect of the rigidity assumption on the behavior of the open loop system is discussed.

Visual lane and obstruction detection system for commercial vehicles

Abstract This paper presents the development of an experimental system for the visual detection of unintended lane-departures of commercial road vehicles. If such event is detected, a warning signal is sent to the driver and a message to the supervisor system. The supervisor can intervene in the motion of the vehicle while it monitors the driver’s activity and the status of the drive stability control system. The edge detection algorithms, which are used to recognize lane geometry, and which are the basis of lane-departure detection, are summarized. The paper presents another experiment in the detection of a three-dimensional object, i.e. an obstruction, in the lane ahead of the vehicles in order to prevent too short safety gaps. The distance can be estimated by processing the stereo images of two cameras.

System for detection and prevention of unintended lane-departure of vehicles

Abstract The paper presents an experimental system for detection and prevention of the unintended lane-departure of road vehicles. The system is realised by a co-operation of a drive stability system and a computer vision system. The detection algorithm does not depend on the type of the lane-boundaries (lane-marks, paved lane-boundary, transition from asphalt to grass) and filters out information marks which do not signalise laneboundaries. If lane-departure is predicted, then intervention to the vehicles motion is applied by unilateral braking. A control algorithm is presented which calculates the necessary torque to bring the vehicle back to the lane.

Grey box identification of vehicle dynamics

Abstract The parametnzation and identification of vehicle vibrating structures are discussed in this paper. The determination of physical parameters from the abstract parameters that represent free coefficients in a transfer function or state space model of the vehicle is investigated for the axle subsystems. The situation when some of the coefficients are known a priori, and also there are certain dependences, possibly nonlinear, among some physical parameters is investigated. Related to this grey box problem, an identification scheme is proposed to obtain the physical parameters from the unconstrained estimates of the abstract parameters of the axle models.