Control of a gyroscopically stabilized robot

Abstract

We have been developing a single wheel, gyroscopically stabilized robot, initially at Carnegie Mellon University. It is a basically single wheel which connected to a spinning flywheel through a two-link manipulator at the wheel bearing. The nature of the system is nonholonomic, nonlinear and underactuated. In this paper, we first develop a dynamic model and decouple the model with respect to the control inputs. We then study the effect of the flywheel dynamics on stabilizing the single wheel robot through simulation and experimental study. Based on understanding of the robot dynamics, we design a linear state feedback controller for the lean angle of the robot. Finally, we discuss the possibility of transferring operator’s manual control directly via learning the mapping between data acquired from the manual control interface and the sensor data from the robot on-board sensors. The experimental results supported the method. The work is significant for understanding the highly coupled dynamics system, and is valuable for developing the automatic control for such a dynamically stabilized robot.