Range estimation from a moving camera: An Immersion and Invariance approach

The paper proposes an original solution to the range identification problem for perspective dynamical systems. The depth of a static point observed by a pinhole camera undergoing a predefined 3-D motion, is estimated from its 2-D projection on the image plane. The proposed nonlinear observer relies on the immersion and invariance (I&I) methodology and offers several advantages over the existing range estimators. The paper also provides an analytical study of nonlinear observability performed with the extended output Jacobian. Extensive simulation experiments illustrate the theory and show the effectiveness of the proposed design.