Closed-Loop Simulation of a Vehicle System with an Artificial Driver

Abstract

In this paper, a model for closed-loop simulation of a driver-vehicle-road system that employs the visual data processing method is suggested. In the model of a human driver, the human capability of learning to negotiate the curved road is simulated using a neural network technique. The learned network extracts an equivalent curvature from the viewing image of the frontal road. This equivalent curvature becomes a control cue for steering action. For computer simulation of the driver-vehicle-road system, a mathematical model of a vehicle with fifteen degrees of freedom is established using the suspension super-element concept, which can handle the three-dimensional motion and suspension kinematics of a vehicle system efficiently. To validate the model of the driver-vehicle-road system, an actual driving rest is performed and the results are compared with those from the computer simulation of the model.