Abstract—Multi agent decision making is a challenging problem in Artificial Intelligence field of research. Coordination between different agents which may be homogeneous or heterogeneous has different complexities. Communication limitation, accessibility to local (non-complete) information, various world models for different agents are some of such problems. Coordination and cooperation have key roles in multi-agent decision making systems. Conclusively, coalition formation has been an attractive and practical approach in multi-agent coordination. An efficient algorithm is required to decide about optimum number of coalition and also number of agents in each coalition. In rescue environment a dynamic algorithm for coalition formation required as situation change. Rescue Simulation Environment is a heterogeneous multi-agent dynamic environment that its aim is to simulate large urban disasters and rescue agent activities and considered as an appropriate platform for associated research. In this paper we propose new algorithms based on Earliest Deadline First scheduling for assemble several rescue teams for various rescue missions. There is also presented team’s disassemble after performance of rescue missions and new rescue team’s assembly due to new situation in the environment. To address a real test case we choose RoboCup Rescue Simulation environment.
Index Terms—coordination; multi-aegnt system; coalition formation; earliest deadline first; RoboCup; Rescue Simulation.
O. Amirghiasvand is with the ECE Department of Azad University of Qazvin, Barajin Ave, Qavin, Iran. (e-mail: o.amirg@gmail.com).
M. A. Sharbafi is with the ECE Department of Azad University of Qazvin, Barajin Ave, Qavin, Iran. (e-mail: Sharbafi@qiau.ac.ir).
Cite: Omid Amir Ghiasvand and Maziar Ahmad Sharbafi, "Using Earliest Deadline First Algorithms for Coalition Formation in Dynamic Time-critical Environment," International Journal of Information and Education Technology vol. 1, no. 2, pp. 120-125, 2011.