Page 59 - Innovations in Intelligent Machines

48 P.B. Sujit et al.
Search Region
100
Performance vs No. of steps
Targets
100
greedy
team theory
80 communication
% target destroyed 60

UAVs 40
20

0
0 0 50 100 150 200
100
No. of steps
(a) (b)
Performance vs sensor radius Performance vs No. of steps
80 100 Team Theory
% Value of target destroyed 70 % Value of target destroyed 60
Greedy
75
80
with communication
65
60
40
55
50
45 20 0
5 10 15 20 25 30 35 40 0 50 100 150 200
Sersor radius No. of steps
(c) (d)
Fig. 2. (a) Search region with UAVs and targets (b) Number of targets destroyed
completely (c) Performance of target value destroyed with variation in sensor range
for the UAVs (d) Average target value destroyed; performance on averaging over 20
different maps

Limited Sensor Range with Full Communication
Here, each UAV has limited sensor range s r but can communicate with all
the other UAVs. Whenever new information is sensed by a UAV, the UAV
broadcasts the information to all the other UAVs. We assume that there are
no communication delays. Hence, all the UAVs have the same information
about the state of the environment at any given time. So, all the UAVs solve
the same LP problem. Moreover, the concept of virtual target does not apply
here, as the i th UAV knows the number of targets present in the neighbours’
sensor region through communication. Similar to the greedy strategy, the UAV
would like to maximize the expected value of the target. The i th UAV solves
the following problem

max C ij x ij (17)
x ij