TY - GEN
T1 - Global Optimal Path Planning for Multi-agent Flocking
T2 - 2019 IEEE Symposium Series on Computational Intelligence, SSCI 2019
AU - Kesireddy, Kesireddy
AU - Shan, Wanliang
AU - Xu, Hao
N1 - Publisher Copyright:
© 2019 IEEE.
PY - 2019/12
Y1 - 2019/12
N2 - Flocking is a multi-objective operation performed by multiple agents in uncertain environments. Objectives of flocking include reaching target for each agent, avoiding collision with obstacles and other agents, as well as maintaining certain pattern among all agents. Multi-objective optimization can be performed in priori methods, posteriori methods and scalarizing methods. Pareto front optimization is the best way to optimize multiple objectives simultaneously. To date, flocking has been performed with summation of objective values. In this paper, Pareto front optimization is adopted for the first time for flocking simulation for multi-agents in uncertain environments. For a team of agents, e.g. rovers, in an uncertain environment, Cooperative Co-Evolutionary Algorithm (CCEA) performs well for both exploration and exploitation. CCEAs coupled with Non-Dominated Sorting Genetic Algorithm II (NSGA-II) and NSGA-III are performed to achieve flocking for multi-agents. A new reward structure is introduced for CCEA. In order to check the reward structure, flocking is performed in different environment, open and closed. In addition, the performances of NSGA-II and NSGA-III are compared for various cases of flocking with different numbers of objectives. Towards the end, the effectiveness of the developed methods is demonstrated through numerical simulations.
AB - Flocking is a multi-objective operation performed by multiple agents in uncertain environments. Objectives of flocking include reaching target for each agent, avoiding collision with obstacles and other agents, as well as maintaining certain pattern among all agents. Multi-objective optimization can be performed in priori methods, posteriori methods and scalarizing methods. Pareto front optimization is the best way to optimize multiple objectives simultaneously. To date, flocking has been performed with summation of objective values. In this paper, Pareto front optimization is adopted for the first time for flocking simulation for multi-agents in uncertain environments. For a team of agents, e.g. rovers, in an uncertain environment, Cooperative Co-Evolutionary Algorithm (CCEA) performs well for both exploration and exploitation. CCEAs coupled with Non-Dominated Sorting Genetic Algorithm II (NSGA-II) and NSGA-III are performed to achieve flocking for multi-agents. A new reward structure is introduced for CCEA. In order to check the reward structure, flocking is performed in different environment, open and closed. In addition, the performances of NSGA-II and NSGA-III are compared for various cases of flocking with different numbers of objectives. Towards the end, the effectiveness of the developed methods is demonstrated through numerical simulations.
KW - NSGA-II
KW - NSGA-III
KW - evolutionary algorithm
KW - flocking
KW - multi-objective optimization
UR - http://www.scopus.com/inward/record.url?scp=85080906839&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85080906839&partnerID=8YFLogxK
U2 - 10.1109/SSCI44817.2019.9002956
DO - 10.1109/SSCI44817.2019.9002956
M3 - Conference contribution
AN - SCOPUS:85080906839
T3 - 2019 IEEE Symposium Series on Computational Intelligence, SSCI 2019
SP - 64
EP - 71
BT - 2019 IEEE Symposium Series on Computational Intelligence, SSCI 2019
PB - Institute of Electrical and Electronics Engineers Inc.
Y2 - 6 December 2019 through 9 December 2019
ER -