Navigation for wheel mobile robot based on EKF with global vision system

In this paper, a new approach to wheel mobile robot navigation with global vision system is proposed. The moving object detection method with GMM(Gaussian Mixture Model) based background modeling is applied to extract the moving robot object from the image captured by a wide angle camera fixed on ceiling. Through a mapping transformation, the position of robot in the real world can be measured and the Extended Kalman Filter is employed to obtain more accurate navigation data. With a series of simulations and experiments, the validity of this approach to wheel mobile robot navigation is verified.