In this study, a genetic-fuzzy control system is used to control a riderless bicycle where control parameters can adapt to the speed change of the bicycle. The equations of motion of a bicycle with constraints of rolling-without-slipping contact condition between wheels and ground are developed. This controller consists of two loops: the inner is a roll-angle-tracking controller which generates steering torque, and the outer is a path-following controller which generates the reference roll angle for the inner loop. The inner loop is controlled by a sliding-mode controller (SMC) on the basis of a linear model obtained from the non-linear one via system identification. By defining a stable sliding surface of error dynamics and an appropriate Lyapunov function, the bicycle can reach the roll-angle reference in a finite time and follow that reference without chattering. The outer loop determines the proper reference roll-angle by using a fuzzy-logic controller (FLC) on the basis of preview distance and direction errors. The robustness of the proposed controller against speed change and external disturbances is verified by simulations. (Author/publisher) For this paper, other papers and posters presented at this Symposium see http://bicycle.tudelft.nl/bmd2010/
Samenvatting