Experimentally Validated Extended Kalman Filter for UAV State Estimation Using Low-Cost Sensors

Abstract

Visually based velocity and position estimations are often used to reduce or remove the dependency of an unmanned aerial vehicle (UAV) on global navigation satellite system signals, which may be unreliable in urban canyons and are unavailable indoors. In this paper, a sensor-fusion algorithm based on an extended Kalman filter is developed for the velocity, position, and attitude estimation of a UAV using low-cost sensors. In particular an inertial measurement unit (IMU) and an optical-flow sensor that includes a sonar module and an additional gyroscope are used. The algorithm is shown experimentally to be able to handle measurements with different sampling rates and missing data, caused by the indoor, low-light conditions. State estimations are compared to a ground-truth pose history obtained with a motion-capture system to show the influence of the optical-flow and sonar measurements on its performance. Additionally, the experimental results demonstrate that the velocity and attitude can be estimated without drift, despite the magnetic distortions typical of indoor environments.