Bayesian Forecasting in Multi-vehicle Search Operations

This paper discusses robust UAV search operations in the context of decisionmaking under uncertainty and presents a new metric called the modifled Bayes Factor (MBF) that encompasses the notion of uncertainty and can be used for robust planning. Various results are shown with the MBF embedded in our planning schemes. This paper also presents a new approach to predict the value of future information by evaluating the most likely future measurements by using Bayesian forecasting, and using these pseudomeasurements to calculate the expected MBF or reduction in variance if additional observations were available. Numerical results are shown that demonstrate the improvement that can be obtained when comparing these robust approaches to nominal ones in a poorly known environment. Key beneflts are also demonstrated with the use of Bayesian forecasting in predicting the most likely locations for improvement in information.