 |
    |
Solution to the Fixed Airbase Problem for Autonomous URAV Site Visitation Sequencing
Amit Agarwal1, Meng-Hiot Lim2, Chan Yee Chew2, Tong Kiang Poo2, Meng Joo Er1, and Yew Kong Leong1
1Intelligent Systems Center, Techno Plaza, Nanyang Technological University Singapore 639798
pg0212208t@ntu.edu.sg
emjer@ntu.edu.sg
leongy.dynamics@stengg.com
2School of Electrical & Electronic Engineering, Nanyang Technological University Singapore 639798
emhlim@ntu.edu.sg
810304085117@ntu.edu.sg
S7931481A@ntu.edu.sg
Abstract. In [1], we modeled the problem of safe site visitation sequencing for an unmanned reconnaissance aerial vehicle (URAV) in graph-theoretic language. Our goal was to uncover several alternative flight paths along which the URAV can fly safely, no more than once, to all sites of interest. We offered a solution to a simpler abstraction of this problem using GAs – if we could not uncover any Hamiltonian cycles we reported maximally long paths between any two vertices. This work reports two orthogonal extensions of [1] – a. In addition to possibly reporting Hamiltonian cycles, the GA reports several alternative long paths that begin at the airbase. b. A randomized version of the gene migration scheme is used. The randomization does not appear to affect the quality of the best plan. However it raises the average quality of the best k plans. We give a modified gene encoding scheme and report simulation results.
LNCS 3103, p. 850 ff.
lncs@springer.de
© Springer-Verlag Berlin Heidelberg 2004