Polarized-Light-Aided Attitude Determination In Integrated Navigation Systems

As a new navigation system, integrated navigation gets more and more focus due to its strong independence,low cost,high accuracy and wide prospect. Polarization information and optical flow are the ways of insects to perceive and process outside optical information. It can be used to estimate the speed,relative depth and attitude angles to detect barriers and avoid obstacle. This dissertation investigates the autonomous , visual-navigation inspired from insects'vision system. The study follows a certain route that the biological principle is the origin, based on which the algorithm description and developed, finally experimental results are analyzed. Using polarized light,optical flow,GPS,SINS as integrated navigation alternative system can get a pretty good performance than any single part does with the help of distributed Kalman filter.Based on all sky polarized light distribution model and navigation principle, a three-channel polarized light measurement is build to polarization information collection. The principles of the polarized light aided attitude determination approaches are described and the equation of the polarized light measurement models of attitude errors are developed. Imaging systems can determine optical flow by tracking the image's features or by computing the ratio of spatial to temporal intensity gradients in the image. Polarization information and optical flow through Kalman filter is introduced to integrated navigation of GPS, SINS, polarized light and optical flow. To increase the performance of observability of the GPS/SINS integrated navigation system, the polarized light and optical flow aided navigation methods are proposed. The results of simulations and estimations of altitude, attitude angles, and velocities of aircraft show high accuracy. The new method of integrated navigation is developed to meet the chanllenges of Mars due to its thin atmosphere , low gravity and a too weak magnetic field for UAVs. With the polarized light and optical flow, a modified visual tracking method is derived, which in SLAM simulations reduces the need for points and allows tracking for extended periods of time without any point registration in Mars 2D planet surface.