tailieunhanh - Innovations in Intelligent Machines 1 - Javaan Singh Chahl et al (Eds) Part 3

Tham khảo tài liệu 'innovations in intelligent machines 1 - javaan singh chahl et al (eds) part 3', kỹ thuật - công nghệ, cơ khí - chế tạo máy phục vụ nhu cầu học tập, nghiên cứu và làm việc hiệu quả | Predicting Operator Capacity for Supervisory Control of Multiple UAVs 29 Table 2. Optimization Parameters Name Unit Value Mission Data includes number of targets - 5-10 targets time on targets and locations UAV speed mi hr 100 UAV Endurance hr 5 UAVs launch location Cartesian 0 0 Cost_per_missed_target target 1500 Cost_of_fuel_per_min min 10 Cost_of_operations_per_min min 1 NT min 32 IT min service rates as well as their corresponding probabilistic distributions are as assumed earlier. We chose the simulated annealing SA technique for heuristic-based optimization. There were several benefits to selecting the SA technique over other optimization techniques. First SA is a technique that is well suited to avoiding local minima a property that is necessary when sub-optimal solutions can exist while searching for the global optimum as is the case in evaluating different mission plans. Also SA introduces randomness such that the technique generates alternative acceptable solutions on different runs hence allowing the system designer to seek alternative optimal designs when initial solutions are not feasible. Two limitations of SA are that problems with many constraints can be difficult to implement and that run times can be long. Our problem however is one of few constraints and hence their implementation was not an issue. Also since optimization takes place in mission planning stages and not in time-critical mission replanning the long run times have a minimal adverse effect. Model Parameters Constraints and Variables. The list of parameters established for the design process is presented in Table 2. We selected generic UAV capabilities that would be exhibited by small-to-medium size UAVs engaged in an ISR mission such as the Hunter or Shadow. Our cost function was discussed previously 5 and Table 3 details the constraints used in our model. Results of Simulation We first investigated the cost-UAV number relationship for the theoretical best case in which the .