Active eye sensing system-predictive filtering for visual tracking

We propose an active eye sensing system (AESS) that tracks and measures the position of a moving object. We take an approach of using binocular vision. It consists of two camera pods and each pod has two degrees of freedom. By gaze control (i.e. tracking the object) and by changing the camera parameters with varying object size and distance from the device, we will get a wide measurement area, reduce the motion blur and sampling error, and improve the accuracy of measurement. In this paper, we report the concept of AESS, system configuration, formulation of the predictive filter that is based on Kalman filter, simulation results of binocular tracking, and basic experimental results