RescueModel: A Multi-Agent Simulation of Bushfire Disaster Management
The RescueModel project is a vehicle for research into multiagent systems, architectures, and strategies. It builds on the theoretical, practical, and experimental base of a decade of beliefs-desires-intentions (BDI) agent systems development. This paper
- PDF / 172,394 Bytes
- 6 Pages / 451 x 677.3 pts Page_size
- 57 Downloads / 166 Views
2
[email protected]
Department of Software Engineering and Computer Science The University of Melbourne, Melbourne, Australia [email protected]
Abstract. The RescueModel project is a vehicle for research into multiagent systems, architectures, and strategies. It builds on the theoretical, practical, and experimental base of a decade of beliefs-desires-intentions (BDI) agent systems development. This paper describes a project that will bring together the environmental richness found usually in large scale military operations research simulations with the architectural richness of agent models often researched in universities. Proposed applications of RescueModel include search and rescue and disaster response studies.
1 Introduction The scenario is large scale bush res and re ghting in the Australian environment. This is currently studied in human decision making and human-in-the-loop (H-I-L) simulation [8]. The RescueModel complements the RoboCup-Rescue simulator [7]. Several re modelling systems exist. They include the Experimental Knowledge Systems Laboratory's Phoenix system [2] for Yellowstone National Park. La Trobe University's networked FireChief program has been used by Australia Defence Science and Technology Organisation's (DSTO's) Information Technology Division to study command and control issues [8]. CSIRO has developed a system called SiroFire [4], a physical model of re perimeter spread. Some aspects of each of these approaches are incorporated into the RescueModel framework [1]. For instance, CSIRO's SiroFire simulator is being integrated as a physical model. However, in SiroFire the cultural features like roads, tracks and railways, and certain natural features such as rivers and creeks are displayed but not used in calculating the re spread. If researchers are interested in collaboration and commitment studies in the context of re ghting tactics for instance, these features become centrally important. Equally important is threedimensional visualisation from the point of view of each agent. Thus a polymorphic representation of agent friendly features is required. In other words, there is P. Stone, T. Balch, and G. Kraetzschmar (Eds.): RoboCup 2000, LNAI 2019, pp. 285-290, 2001. c Springer-Verlag Berlin Heidelberg 2001
286
Gary Au et al.
a need for a representation containing polygon information for scene rendering, and also feature labels which agents can reason about. The main drawback to many of the above re approaches is that the representation of the environment is not accessible to agents in a structured way. In operational simulations of other domains, such as military operations, it has proved bene cial to clearly de ne activity in the simulation in terms of an agent model. Such agent systems have proven valuable for operational simulations in air combat [11]. The simulation infrastructure of RescueModel is a variation of a model developed by the Defence Science and Technology Organisation of Australia, named BattleModel. BattleModel was developed as a simulatio
Data Loading...
