Range and Motion Estimation of a Monocular Camera Using Static and Moving Objects

We propose a method of estimating the motion of a monocular camera looking at moving objects and their range. Unlike the previous studies where the camera and object motion should be constrained in estimating structure and motion (SaM) of moving objects, the proposed method do not require those constraints even though only a monocular camera is used. By first arranging the SaM dynamics in terms of the measurable states, we design robust nonlinear observers in a sequential way for both static (stationary) and dynamic (moving) objects. Through the combination of these estimates obtained by nonlinear observers, the reconstruction of the 3-D structure of the dynamic objects can be achieved using just 2-D images of a monocular camera. Simulations are performed in the case of changing camera and object velocities, such that the advantages of the proposed method can be clearly demonstrated.