Active Control for a Riderless Bicycle

Abstract

The primary objective of this thesis was to develop an active control system for maintaining the balance of a riderless bicycle traveling at a constant velocity. In past studies, a popular method for controlling the posture, or upright stability, of the bicycle has been the actuation of the front fork. Most successful experimental implementations of this actuation method have utilized control of the steering torque. Maintaining bicycle stability through the direct actuation of the steering angle is a control method that has not been as well explored. The focus of this study was thus to examine whether directly controlling the steering angle could be effective in maintaining bicycle stability in both simulations and experiments.

An idealized model of the bicycle system was first utilized to develop an active control system that could maintain the posture of a bicycle in simulations. An ex- tension of this control system that would enable the bicycle to maintain its posture while tracking a predefined trajectory was also developed and validated through further simulations. The idealized model was then modified to more closely emulate the experimental bicycle system by accounting for the performance of the actuator, a servo motor with an inherent time delay, and for the low pass filter used to process the data from the chosen tilt sensor. Several controllers of varying complexity were then designed and evaluated to maximize the robustness and performance of the final system, and the most viable ones were tested on the bicycle. Simulations demonstrated that although maintaining stability was possible in theory, the delays in the experimental system imposed significant limitations on the robustness of the control system. In the experimental trials, the riderless bicycle exhibited improved stability, but the limitations proved to be too restrictive in practice, and lasting stability was not observed.