Page 38 - Innovations in Intelligent Machines
P. 38

26     M.L. Cummings et al.
                                       C=cost of fuel total UAV distance
                                                     ∗
                                           +cost per missed target # of missed targets
                                                                ∗
                                                                 ∗
                                           + operation cost per time total time             (5)
                              In order to maximize performance, the cost function should be minimized
                           by finding the optimal values for the variables in the cost equation. However,
                           the variables in the cost equation are themselves dependent on the number of
                           UAVs and the specific paths planned for those UAVs. One way to minimize the
                           cost function is to hold the number of UAVs variable constant at some initial
                           value and to vary the mission routes (individual routes for all the UAVs) until
                           a mission plan with minimum cost is found. We then select a new setting for
                           the number of UAVs variable and repeat the process of varying the mission
                           plan in order to minimize the cost. After iterating through all the possible
                           values for the number of UAVs, the number of UAVs with the least cost and
                           the corresponding optimized mission plan are then the settings that minimize
                           the cost equation. As the number of UAVs is increased, new routing will be
                           required to minimize the cost function. Thus, the paths, which determine time
                           of flight, are a function of number of UAVs.
                              Moreover, if a target is missed, then there is an additional, more significant
                           cost. When the number of UAVs planned is too low, the number of missed
                           targets increases and hence the cost is high. When the number of UAVs is
                           excessive, more UAVs are used than required and thus additional, unnecessary
                           costs are incurred. We therefore expect the lowest cost to be somewhere in
                           between those two extremities, and that the shape of the cost curve is therefore
                                          1
                           concave upwards (Figure 8). The profile in Figure 8 does not include the effect
                           of wait times, and it does not take into account the interaction between the
                           vehicles and the human operator.

                                            Too many
                                           missed targets



                                      Mission Plan Cost            # of UAVs with  UAVs
                                                                             Too many




                                                                   minimum cost

                                                     # of UAVs
                                   Fig. 8. Mission Plan Costs as a Function of Number of UAVs


                            1
                              Note that this claim is dependent on the assumption that the UAVs indepen-
                             dently perform tasks.
   33   34   35   36   37   38   39   40   41   42   43