Closed-Form Solution for Lost-In-Space Visual Navigation Problem

摘要

The paper describes a new closed-form solution for the lost-in-space visual navigation problem. Focal plane images of several known beacons are used to determine both position and attitude of a sensor equipped with position-sensing diodes. A preliminary set of possible solutions is obtained without any initial guesses using any three beacons. These solutions employ a closed-form method for finding sensor-to-beacon distances based on their lines of sight. Instead of providing these solutions in terms of polynomial roots, the paper for the first time derives these solutions explicitly and includes a complete error analysis. Using additional beacons, the possible three-beacon solutions are refined, extraneous solutions are discarded, and the best solution is identified via the Gauss-Newton method. The paper also includes numerical examples and a statistical validation of the proposed method.