Abstract
In this paper, we present a linearlized system for a motorcycle, for which its stability and model analysis are performed. The steering torque to perform lane change is obtained by applying final-output control with error learning. The equation of motion for the motorcycle was obtained using multi-body dynamics analysis assuming that the motorcycle consists of four rigid parts with nine degrees of freedom. We linearized the equation of motion around a point where the velocity is constant when runnning straight. The linearized model were separated as the capsize, weave and wobble modes. As a result of the frequency response analysis, it turned out that the gain of the weave mode is largest at the speed of 30km/h. From comparison between before learning and after, it was verified that the error learning improves the state of the motorcycle concerning the weave mode in the lane change.
