Integrated Planning, Decision-Making, and Weather Modeling for UAS Navigating Complex Weather.

Aircraft of all types, and especially small UAS, are significantly affected by atmospheric motion. Employing numerical weather models and trajectory planning algorithms that reason over model uncertainty can allow an aircraft to safely and efficiently traverse a complex, uncertain environment. However, even paths robust to a priori uncertainty may be inferior to trajectories planned using an environmental model refined using in situ observations. This work develops a dynamic data driven applications system (DDDAS) architecture that uses the aircraft as a sensor to the environmental state and updates a model of the wind field model through trajectory execution. A Monte Carlo Rapidly-Exploring Random Tree (MCRRT) algorithm plans a set of probabilistically safe paths and predicts the distribution of their cost. Decision-making at the tasking level directs aircraft on paths which are possibly suboptimal with respect to a single mission in order to sample the environment and update the model. This tasking reconfigures the observations gathered to target portions of the environment relevant to mission execution. Initial simulations show that this approach is able to reduce error in the modeled environment.