No abstract

Coupled operator-multiple vehicle systems are modelled in a unified framework using probabilistic graphs to yield a methodology for analyzing semi-autonomous systems. The framework uses conditional probabilistic dependencies between all elements, leading to a Bayesian network (BN) with probabilistic evaluation capability. Vehicle attitude/navigation states and target/classification states can be evaluated using nonlinear estimators such as the EKF, Multiple Model filter, information filter, or other approaches. Discrete operator decisions are being modeled as Bayesian network blocks, with conditional dependencies on the vehicle and tracking estimators. Initial decision models use combinations of softmax and discrete probability distributions.

Of the methods developed for Optimal Task Allocation, Mixed Integer Linear Programming (MILP) techniques are some of the most predominant. A new method, presented in this paper, is able to produce identical optimal solutions to the MILP techniques but in computation times orders of magnitude faster than MILP. This new method, referred to as G*TA, uses a minimum spanning forest algorithm to generate optimistic predictive costs in an A* framework, and a greedy approximation method to create upper bound estimates. A second new method which combines the G*TA and MILP methods, referred to as G*MILP, is also presented for its scaling potential. This combined method uses G*TA to solve a series of sub-problems and the final optimal task allocation is handled through MILP. All of these methods are compared and validated though a large series of real time tests using the Cornell RoboFlag testbed, a multi-robot, highly dynamic test environment.