Abstract

An autonomous avoidance system for the use in UAVs in controlled and uncontrolled airspace has to be able to ensure that a UAV is able to fulfil the 'Sense-and-Avoid' requirement with an equivalent level of safety to inhabited aircraft in which the pilot is able to monitor the direct surroundings of the aircraft and has direct control over the aircraft systems. The focus of this paper is on the avoidance functionality of the system. For this functionality a decision making architecture is introduced which is able to determine whether an intervention is required and then define an avoidance strategy and generate an avoidance manoeuvre. A modular approach ensures that only those parts of the system which are critical to the safety of the aircraft need to be certified as such. Depending on the scenario the 'Sense-and-Avoid ' requirement is fulfilled using a hybrid approach of a reactive and/or deliberate avoidance manoeuvre generation algorithm. The deliberate part ensures sufficient separation between aircraft such that a potential conflict is avoided in the first place. The reactive function acts as a safety net to avoid an imminent collision hazard. Due to the different requirements for each function two types of algorithms are used. The separation function is based on a deliberate short-term path search. The safety net uses a direct command reactive algorithm. Both algorithms are able to handle constraints with respect to airspace, aircraft performance and multiple intruding aircraft. 1.