8.9.1 Research on cooperative multiagent systems that can plan, problem solve, learn, and make decisions in a partially unpredictable environments is particularly of interest to military strategists. In such contexts, important new information about something other than the current goal can arrive at unexpected times or found in unexpected contexts, and there is often not sufficient time for deliberation. (Sloman, Logan, 1999, pg. 72) Perhaps the biggest challenge facing the military in agent research is how to adapt to different cultures whose logic may not quite match those who are directing the programming. In order to deal with this, some of the most advanced research in motivational and emotional requirements for intelligent agents has been sponsored by the military. (Sloman, 1997, pp. 166-208). Thus the focus of Reactive Agents, where detection of internal and external conditions immediately generates new internal or external responses, in turn trigger new reactions. This kind of architecture requires a large amount of stored knowledge, including which actions are possible or relevant under certain circumstances and what the various effects of certain actions are in those circumstances.

8.9.2 The framework used most for military experiments with agents is the SIM_AGENT toolkit. It allows multiple agents to be run and controls their communication with each other and with the physical simulation of the battlefield. Simulated battlefield commanders or simulated antiterrorist strategists may have to detect and handle conflicts between protecting civilians and capturing opponents. These agents have already been deployed to control tanks in ground battle simulations used in military training. The tactical behaviour is governed by the terrain over which the tanks are moving and the beliefs of the enemy. The hierarchical structure of the SIM_AGENT system remarkably resembles that of the military. High level commanders are given objectives that are used to produce lower-level objectives for their subordinates.

8.9.3 Information flows both up and down the command chain and agents need to cooperate with their peers to achieve the overall goal set by their commander. This natural decomposition of the problem allows higher-level agents to work on long-term plans while the individual tank agents carry out orders designed to achieve more immediate objectives. (Baxter, Hepplewhite, 1999. pg. 74) [7] [top] Notes:

7. The Birmingham Poplog has a directory of information and file lists on SIM_AGENTS and its supporting libraries: ftp:ftp.cs.bham.ac.uk/pub/dist/poplog Poplog is available from Integral Solutions Ltd.: www.isl.co.uk [back]