Autonomous Search, Detection, Geolocation, and Payload Drop: NKRAFA’s Solution for the Autonomous Aerial Vehicle Challenge 2023

Navaminda Kasatriyadhiraj Royal Air Force Academy (NKRAFA) won the Autonomous Aerial Vehicle Challenge (AAVC) 2023. This paper describes the architecture, design, implementation, and testing of an unmanned aircraft system that ultimately won the challenge. The mission of the challenge was to search for a mannequin target, acquire a geolocated coordinate of the target, and then drop a payload as close to the target as possible. The scoring was based on the level of autonomy in each phase of the mission. During the competition, NKRAFA was able to fly the mission autonomously from takeoff to landing. NKRAFA earned a full score and only missed the extra points. The mission was completed within 8[Formula: see text]min out of the 25[Formula: see text]min allotted time. Our multirotor aircraft was operated according to the two-internal pilot operational concept — where the first internal pilot will be responsible for the flight aspect of the operation, while the second internal pilot will be responsible for the mission aspect of the operation. The hardware architecture also reflected this concept. There were two onboard computers and two ground control stations. Our mission software components included a YOLO v4 object detector, the target candidate filter, the Extended Kalman Filter states estimator, the geolocation, and the mission manager. The paper will also discuss emergency procedures, our experiences from flight testing, and the results of the competition.