Modeling of Vehicle based on Visual Feedback Control by Closed-loop Identification

Abstract

This paper considers the usefulness of modeling of vehicle based on the visual feedback system by closedloop identification. In this paper, we consider a radio control car (RC-car) that is equipped with an onboard camera. A steering of RC-car is the input and a feature which can be calculated from a set of feature points on an image is the output. This research goal is to design the two models for the visual feedback controller. The RC-car runs not only on straight but any curves. For this reason, we need to measure output data on curve driving with random input in order to design the curve model using input and output data sets. The Modeling based on system identification has to use the persistently exciting signal, the driving of RC-car with random input is however difficult. It is necessary to get input-output data to the controller. Therefore, we have to use a closed-loop identification for taking data of identification. We make a model of RC-car by closed-loop identifications and we investigate validity.