An Integrated and Vision Aided GPS/INS Navigation System for Ultra-low-cost MAVs
Today's MAV navigation systems take advantage of GPS in combination with an IMU to derive velocity and position as well as attitude and heading. Ultra-low-cost systems (<500$) for unstable flight vehicles, with their restricted hardware abilities and low measurement quality, use GPS information to determine the position and inertial measurements for stabilization control only. This paper presents an integrated closed-loop navigation system with ultralow-cost MEMS inertial sensors using a complementary Kalman filter approach, which fuses GPS and IMU data in a loosely-coupled-system to improve position and velocity information. Therefore, sophisticated approaches from highend UAVs have been adapted for usage in very cheap MAV navigation systems. As further improvement, this paper introduces additional aiding by computer vision and a smooth operating mode switching concept. This concept allows seamless switching between different modes of operation, which are selected depending on the availability of visual information and GPS data. Correct attitude information needed for stabilization control can be guaranteed also in case of permanent GPS and/or visual data loss. Implemented on a tiltwing MAV, this solution is also capable of dealing with both flight modes. The approach is evaluated by simulation, based on empirically captured real-world data.
Felix Gathmann, Christian Dernehl, Dominik Franke and Stefan Kowalewski