Abstract
In this paper, we consider visual feedback stabilization which tolerates small camera misalignment. Specifically, a balancing task with a cart-pendulum system using camera image is examined. Such a task is known to rely heavily on the detection of the vertical direction and the angle measurement error due to the camera misalignment could be fatal for stabilization. From a mathematical model of the measurement error, the effect of the misalignment is naturally represented by affine perturbation to the coefficient matrix of the output equation. Motivated by this fact, a special type of robust dynamic output feedback stabilization against polytopic uncertainty is investigated. By solving the related BMI, one can design a controller which tolerates the camera misalignment to some extent. The result is verified via experiments.
