Obstacle detection and collision avoidance method based on optical systems

• Autor(a) principal: Silva, Ana Beatriz Botelho da
• Data de Publicação: 2017
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10400.6/7951
• topic: Aeronautic; Air Traffic; Collision Avoidance System; Computational Algorithm; Optical Detection; Domínio/Área Científica::Engenharia e Tecnologia::Engenharia Aeronáutica

The development of a new collision avoidance method, which can detect and calculate the necessary changes to prevent imminent accident, is the focal interest of this work. In aviation, the risk of collision is a delicate and important subject, which merits the right approach. With the continuing growth of air traffic and the introduction of RPASs (Remotely Piloted Aircraft System), it is necessary to find better solutions and develop new systems to keep the control of the airspace. In this work, the main objective is to obtain a complete and functional computational algorithm, which could be included in an obstacle detection and avoidance system. Its unique feature of optical detection makes it mostly appropriated for RPASs. The application of Optical Techniques is mostly used in aircrafts to detect objects under them [1] or even to prevent a collision with terrain [2]. Some technologies also use optic flow sensors to detect and prevent collisions [3, 4]. In this case, the optical system will be used to detect obstacles in front of the aircraft. The detection of an obstacle will be performed by the two infrared cameras strategically positioned in the aircraft. The objectives to accomplish with this method are: capable of dealing with collision detection characteristics; in case of detecting a possible threat of collision, describing the safe zone as the area outside a conflict cone; assessing if the threat of collision previously detected is real; in case the danger is real, changing the aircraft’s trajectory by altering one or more flight characteristics. To achieve the most efficient method possible some theoretical methods were explored, like the Convex Hull Method, which is a simple geometrical method, and a variation method based on differential equations. With the aim of testing the algorithm in different situations, a total of six possible cases were generated. All the results showed coherence and efficiency, which confirms the success of this computational algorithm as a detection and collision avoidance method.