This paper presents an approach to visually guide an unmanned aerial vehicle (UAV) through narrow passages. Once a mission objective has been set by the UAV operator, the implemented system operates autonomously and without the need for a data link to the operator. This multi-disciplinary approach comprises a vision-based gate detection, a camera misalignment compensation, a geo-referenced gate localization, as well as autonomous plan adaption and plan execution. From real outdoor flights, the gate recognition algorithms were validated using image sequences and navigation data acquired onboard the helicopter. The overall approach is tested successfully in an autonomous flight test. As a result, the helicopter UAV is able to fly through a narrow gate without collision. The test scenarios show that without a online correction of the estimated gate position, the vehicle would have collided with the gate posts. The onboard replanning capabilities require a low computational resources as these run at constant runtime. Moreover, the reactive planning and a state chart-based modeling of the gate detection sensor handling and the gate passing behavior decompose this complex mission task into manageable phases. The presented approach requires a-priori knowledge about expected obstacles. Hence, upcoming research is going to extend the presented algorithm to handle a gate, window, or other small passages where fewer attributes are known apriori. Thus, future research steps need to relax the assumption of a specific type of narrow passage, e.g. the gates presented in this work.
Vision-based target recognition and autonomous flights through obstacle arches with a small UAV
Sichtbasierte Zielerkennung und autonomer Flug durch Hindernisbögen mit einem kleinen unbemannten Luftfahrzeug
AHS International, Annual Forum, 65 ; 2274-2283
2009
10 Seiten, 16 Bilder, 21 Quellen
Conference paper
English
Vision-Based Target Recognition and Autonomous Flights Through Obstacle Arches with a Small UAV
British Library Conference Proceedings | 2009
|Autonomous vision-based helicopter flights through obstacle gates
Tema Archive | 2010
|Computer vision based obstacle detection and target tracking for autonomous vehicles
DOAJ | 2021
|Autonomous UAV obstacle avoidance using machine learning from piloted UAV flights
European Patent Office | 2020
|Autonomous UAV obstacle avoidance using machine learning from piloted UAV flights
European Patent Office | 2021
|