Vision-based target pursuit by autonomous UAVs

点击翻译按钮获取中文摘要

The proposed research program employs a flying quadcopter Unmanned Aerial Vehicle (UAV), widely used for application such as aerial photography and remote surveillance, and develops the novel capability of employing on-board vision sensors to autonomously detect and follow or intercept a second flying quadcopter UAV. Example applications of this technology include protecting sensitive areas such as government buildings or military installations from a weaponized UAV attack, or guarding private events from unauthorized aerial photography and videography by the media. The research work involves a mix of advanced theory in aided inertial navigation, feedback control and computer vision and practical issues in coding, sensor calibration, parameter tuning and experimental flight testing. This makes it an excellent platform for training the next generation of highly skilled engineers, who are critical for the success of Canada's high-tech sector and increasingly important robotics industry. The algorithms and knowledge developed for this project will be extended and applied to two more complex applications relevant for Canadians. The first extension is autonomously detecting and chasing away birds from pre-designated areas. This technology would be of great interest to airport authorities to prevent dangerous engine bird strikes, oil sands companies to keep birds from landing on tailings ponds, and farmers protecting their crop fields. The second extension is chasing cars for the police. High-speed vehicle chases are known to be extremely dangerous for the drivers, other cars and pedestrians. Because a quadcopter is capable of both low and high-speed flight and nimble maneuvering, it would be able to easily and safely follow a fleeing vehicle while reporting its position to law enforcement.