Autonomous VTOL software for Intelligent and Adaptive Noncooperative landing (AVIAN)

We propose an expansion of AVIAN, a system designed for the purpose of helping VTOL UAS analyze noncooperative naval environments to make predictions and decisions regarding autonomous landing. AVIAN combines various sensor data inputs to model the environment for the intelligent analysis of potential landing sites. This analysis uses literature-backed computer vision algorithms to identify and rate the quality of each potential site. Landing sites are analyzed in a coarse-to-fine manner, meaning poor-quality sites are swiftly pruned out of consideration, saving limited UAS resources. This approach allows for higher-quality sites to be analyzed with computationally expensive operations that would otherwise lag the decision-making process. AVIAN uses this site analysis information along with mission- specific criteria such as stealth requirements to intelligently determine the most appropriate landing site. Upon selecting a site, AVIAN continuously updates its representation of the environment to ensure it can safely and stealthily approach the landing site. Upon reaching a landing site and ensuring that it is of high quality, AVIAN works to safely land, using IMU and camera data to ensure a proper landing. AVIAN then uses its stored landing information and new sensor information to attempt to safely leave while remaining covert.