Small UAV autonomous localization based on multiple sensors fusion

Abstract

Autonomous localization is a challenge for small UAVs in the environment where GPS signal is unavailable. In this paper, we propose a novel autonomous localization algorithm for small UAVs without map construction. In the proposed method, the information of onboard camera, pressure sensor and orientation sensor are integrated together based on Kalman filter model under the visual odometry framework. The algorithm is realized in Robot Operating System(ROS) and is tested based on Gazebo. Experimental results show the localization accuracy of the proposed algorithm under different flight conditions and flight status. It is verified that the method extends the environmental adaptability of existing visual odometry methods.