Decision-Making Related to UAV Control Under Uncertainty

The article is devoted to the decision-making onboard of the unmanned aerial vehicles (UAVs) under complex or, in some cases, even uncertain operating conditions, for example, automatic and even unplanned landing on improvised sites. We have proposed applying decision-making to specific tasks, e.g., decisions to “land” or to “continue flight” within the “Game Against Nature” framework. In addition, decisions are made, while a UAV performs an autonomous flight, which is why various scenarios should be taken into account, based on uncertainty of a current situation. To identify nature’s strategies, situation awareness method has been used in this work to assess the state of the current situation and to predict the consequences of decisions made and controls implemented. The current technology for situation awareness has been implemented, based on the use of databases built in advance and containing information about possible objects of interest, including area maps, knowledge bases containing products that describe cause-and-effect relationships between objects (phenomena, processes), as well as a set of models describing current statuses. The UAV and its subsystems that safeguard efficient task execution—namely, the UAV performance, including range, speed, and maneuverability under current conditions; the operating spectral range of the surveillance system, etc. The environment, e.g., time of day, weather conditions, etc., includes effects of destabilizing factors. Objects of interest specified for a task assigned. We have proved that making decisions about landing or continuing flight through implementing onboard algorithms implies taking into account possible nature’s strategies, the status of the UAV, and the expected pessimism index of the Hurwitz criterion.