Velocity and Path Reconstruction of a Moving Object Using a Moving Camera

Integral concurrent learning based observers are developed that use velocity measurements of a monocular camera and pixel coordinates of tracked features to simultaneously estimate the path of a moving monocular camera, the distance to stationary features, as well as the structure and motion of rigid moving objects features in real time. A Lyapunov-based stability analysis is provided to show the structure and motion observer estimation errors for the moving objects globally exponentially converge to an ultimate bound or zero if the object has a constant velocity.