Vision-Based Nonlinear Feedback Control of a Ball on Ball System With a Programmable Logic Controller

This paper shows how to implement a vision-based control for a Ball on Ball system. The control task is to stabilize a ball on top of another ball in the unstable upper rest position. In order to stabilize the upper ball, the position of the upper ball has to be measured. This is done optionally by using two triangulation sensors or an industrial camera. To stabilize the upper ball in the unstable upper rest position, a PD controller, a linear quadratic regulator, and a flatness-based nonlinear controller are designed. By using these controllers and the designed image processing, which are implemented entirely on a standard industrial programmable logic controller, it is possible to stabilize the upper ball with both measurement types. The disturbance rejection of the system is shown, and the resulting RMS position error when stabilizing the upper ball is 1.8 mrad.