Navigation of a Low Flying VTOL Aircraft With the Help of a Downwards Pointing Camera

Abstract

At the German Aerospace Center (DLR) Institute of Flight Systems a vision system is developed that provides information about the movement of the camera relative to the scene. This system is integrated in the available Autonomous Rotorcraft Test bed for Intelligent Systems (ARTIS) operated by the DLR to test the performance under real world circumstances. A downwards pointing camera is used to track a predefined pattern or arbitrary features on a grass surface. The tracking is based on classic image processing techniques like thresholding and optical flow algorithms. The positions of the acquired features are fed into a non-linear filter that estimates the relative position and velocity between the camera and the environment in all 6 degrees of freedom. Using the filter has the advantage of being able to use the history of movements and not only to relay on a one or two image prediction of the position. That makes it much more robust against pixel noise and uncertainties in the calculation of the optical flow. A simulation is built up to test the system. The conducted simulations reveal a sensitivity of the estimations of the relative position to the initialization of the filter and a good performance of the estimation of the rotation angles.