Towards autonomous tracking and landing on moving target

摘要

The battery capacity of Unmanned Aerial Vehicle (UAV) is the main limitation, but with the rapid growth of UAV deployment in both military and civilian application, there is an urgent need to development the reliable and automated landing procedure. This paper is aim to propose a basic framework for autonomous landing on moving target for the Vertical Take-off and Landing (VTOL) UAVs. The VTOL vehicle is assumed to equip with a Global Navigation Satellite System (GNSS) system and a stereo vision system which could generate the point cloud within 20 meters. We applied a particle filter based Visual Servo in the UAV vision system to detect and track the moving the target at real time. We also combine the inertial measurement unit (IMU) data with the stereo vision based visual odometry to make the relative accurate pose estimation. The relative position, orientation and velocity to the landing area on the moving carrier is obtained by a modified optical flow method. The control method used in this framework combined tracking and approaching base on the range distance. We has applied our proposed framework on both simulation and landing task on a moving vehicle, and the result shows the efficiency and extended ability of our framework.