Simulating Fighter Pilots
3. Agent Research and Development
3.8. Civilian Modelling
Complex terrain includes complex physical terrain, complex human terrain and complex informational terrain. Current military simulations and wargames rarely model complex human terrain and there are even fewer examples that include com- plex informational terrain and as combat within complex terrain becomes the norm civilian modelling becomes more important [29]. To represent civilians embedded
Figure 4. The Codarra Avatar was fitted with a HP IPAQ run- ning a JACK agent connected serially between the radio modem and the flight control system. The agent provide autonomous con- trol of the UAV: monitoring the aircraft state via the data sampled from the flight control system and controlling the aircraft through waypoints fed into the flight control system.
within a complex human and informational terrain a multi agent paradigm was selected as being the most appropriate due to the ability to incorporate different aspects of human behaviour and interaction in a visible structured manner.
Agents as members of social networks moving through their city from home to work is the first step this project has taken. This small step has none the less pro- vided enough complexity to simulate disease spread through the population with results that replicate those provided by mathematical models such as SIR [28].
Initial extensions to these agents will increase the number of activities they un- dertake and allow reasoned responses to changes in their environments. Follow on research will explore emotion, motivation and the effect of cultural backgrounds on behavioural representation.
Figure 5. A screenshot from the Human Agent Virtual Environ- ment (HAVE), a close air support (CAS) simulation. The design of the agent-environment interaction was motivated and inspired by the theory of affordances from ecological psychology; the study of how humans and animals interact with their environments.
4. Conclusions
The impact of intelligent agents within the DSTO can be measured in the suc- cessful deployment of a number of innovative systems, the associated productivity improvements resulting directly from the technology and indirectly from the asso- ciated adoption of improved software engineering practices, and the quality of the research that has been fostered in and around the development. In summarising the advantages of agents it is worth first cautioning that the particular adopted technology matched neatly with the requirements of defence science and though the application development has been successful it has not been without risk, steep learning curves, and cost. As Wooldridge and Jennings caution “There are a num- ber of good reasons for supposing that agent technologies will enhance the ability of software engineers to develop complex distributed applications” but agents are not a magical problem solving paradigm [27].
By creating an environment where tactics and standard operating procedures are more explicitly represented it has been possible to improve the interactions between analysts and military personnel. The abstract graphical representation of plans provides the flight crews with visibility into the simulation increasing their confidence in the modelling. It allows tighter faster validation of tactical plans by inspection of crews and therefore improves the confidence that analysts have in the models.
Transitioning to agent technology shifted the focus from a systems view of military operations to a human centred view. This allowed problems to be viewed in radically different ways, for new research threads to be explored and for col- laboration with psychologists, physiologists, and human factors experts to explore modelling options.
The acquisition of intelligent agent technology has resulted in large gains in productivity. Problems that were previously intractable have been opened up and explored while other problems are now addressed more effectively and efficiently.
The particular implementations and technologies described here are clearly not suitable for all problems in all domains and care must be exercised when selecting novel technologies.
The particular domain and challenges facing the defence science meshed ide- ally with the adopted agent approaches and the significant risks associated with new technologies were mitigated through iterative development, careful tool ac- quisition and a firm research base. The success of a decade of intelligent agent research has resulted in quality outcomes for the Australian Defence Force, both in timeliness and quality, and an internationally recognised research program for DSTO.
References
[1] D. McIlroy, B. Smith, C. Heinze, and M. Turner “Air Defence Operational Analysis Using the SWARMM Model,” inProc. of Asia Pacific Operations Research Sympo- sium, 1997.
[2] C. Heinze, S. Goss, T. Josefsson, K. Bennett, S. Waugh, I. Lloyd, G. Murray and J. Oldfield, “Interchanging Agents and Humans in Military Simulation,” in Proc.
of Thirteenth Innovative Applications of Artificial Intelligence Conference, Deployed Application Case Study Paper, Seattle, Washington, 2001.
[3] G. Tidhar, C. Heinze, S. Goss, G. Murray, D. Appla, and I. Lloyd, “Using Intelligent Agents in Military Simulation or “Using Agents Intelligently”,” inProc. of Eleventh Innovative Applications of Artificial Intelligence Conference, Deployed Application Case Study Paper, Orlando, Florida, 1999.
[4] M. Tambe, R. M. Jones, J. E. Laird, P. S. Rosenbloom, and K. Schwamb, “Building Believable Agents for Simulation Environments: Extended Abstract,” in Collected Papers of the SOAR/IFOR Project, Information Sciences Institute, University of Southern California, pp. 78-81. Marina del Ray, CA, 1994.
[5] M. Tambe, K. Schwamb, and K. S. Rosenbloom, “Building intelligent pilots for simu- lated rotary wing aircraft,” inProc. of the Fifth Conference on Computer Generated Forces and Behavioral Representation, pp. 39-44, 1995.
[6] D. McIlroy, C. Heinze, D. Appla, P. Busetta, G. Tidhar, and A. Rao, “Towards Credible Computer Generated Forces,” inProc. of Second International Simulation Tech-nology and Training Conference, (SimTecT ’97), Melbourne, Australia, 1997.
[7] C. Heinze, B. Smith, and M. Cross, “Thinking Quickly: Agents for Modeling Air Warfare,” inProc. of Australian Joint Conference on Artificial Intelligence, AI ’98, Brisbane, Australia, 1998.
[8] R. L. Shaw, Fighter Combat - Tactics and Maneuvering. 6th edition, US Naval In- stitute Press, 1985.
[9] M. Bratman, Intentions, Plans, and Practical Reason. Harvard University Press, Boston, MA, 1987.
[10] M. d’Inverno, D. Kinny, M. Luck, and M. Wooldridge, “A formal specification of dMARS,” in M. P. Singh, A. Rao, and M. J. Wooldridge, editors,Intelligent Agents IV (LNAI Volume 1365), Berlin, Germany: Spinger-Verlag, 1997, pp. 155-176.
[11] M. P. Georgeff, A. L. Lansky, “Procedural Knowledge,” inProc. of the IEEE Special Issue on Knowledge Representation, vol. 74, pp. 1383-1398, 1986.
[12] D. McIlroy and C. Heinze, “Air Combat Tactics in the Smart Whole AiR Mission Model,” inProc. of First International Simulation Technology and Training Confer- ence, (SimTecT ’96), Melbourne, Australia, 1996.
[13] N. Howden, R. Ronnquist, A. Hodgson, and A. Lucas, “JACK Intelligent Agents:
Summary of an Infrastructure,” in Proc. of the 5th International Conference on Autonomous Agents, 2001.
[14] G. Tidhar, M. Selvestrel, and C. Heinze, “Modelling Teams and Team Tactics in Whole Air Mission Modelling,” inProc. of Eighth International Conference on In- dustrial and Engineering Applications of Artificial Intelligence and Expert Systems, (iea-aie ’95), Melbourne, Australia, 1995.
[15] J. R. Boyd,A Discourse on Wining and Losing. Unpublished set of briefing slides available at Air University Library, Maxwell Air Force Base, Alabama, 1987.
[16] G. Tidhar, “Organization-Oriented Systems: Theory and Practice,” in Department of Computer Science and Software Engineering, 1999, University of Melbourne: Mel- bourne, p. 278.
[17] R. Hodgson, R. Ronnquist, and P. Busetta, “Specification of Coordinated Agent Behavior (The SimpleTeam Approach),” inWorkshop on Team Behavior and Plan Recognition, International Joint Conference on Artificial Intelligence, Sweden, 1999.
[18] C. Heinze, S. Goss, and A. Pearce, “Plan Recognition in Military Simulation: Incor- porating Machine Learning with Intelligent Agents,” inProc. of Workshop on Team Behavior and Plan Recognition IJCAI’ 99, Stockholm, Sweden, 1999.
[19] G. Booch, J. Rumbaugh, and I. Jacobsen,The Unified Language User Guise.Read- ing, MA: Addison Wesley, 1999.
[20] J. Odell, H. V. D. Parunak, and B. Bauer, “Extending UML for Agents,” inProc. of the Agent-Oriented Information Systems Workshop at the 17th National conference on Artificial Intelligence, 2000.
[21] C. Heinze, M. Papasimeon, and S. Goss, “Specifying Agent Behaviour With Use Cases,” inProc. of Pacific Rim Workshop on Multi-Agents, 2000.
[22] M. Papasimeon and C. Heinze, “Extensions to the UML for Designing Jack Agents,”
in Proc. of the Australian Software Engineering Conference (ASWEC), Canberra, Australia, 2001.
[23] G. S. Halford, W. H. Wilson, and S. Phillips, “Processing capacity defined by rela- tional complexity: Implications for comparative, developmental, and cognitive psy- chology,” accepted as target article: Behavioral and Brain Sciences.
[24] C. Heinze and S. Goss, “Human Performance Modelling in a BDI Agent System,”
inProc. of OZCHI, Sydney, Australia, 2000.
[25] R. M. Jones, J. E. Laird, “Constraints on the design of a high-level model of cogni- tion,” inProc. of Nineteenth Annual Conference of Cognitive Science, pp. 124-132, 1997.
[26] M. Papasimeon, A. Pearce, S. Goss, C. Heinze and T. Patterson, “The Human Agent Virtual Environment,” inProc. of 2007 Joint Conference on Autonomous Agents and Multi-Agent Systems, Honolulu, Hawaii, USA, 2007.
[27] M. Wooldridge and N. Jennings, “Pitfalls of Agent Oriented Development,” inProc.
of the 2nd International Conference on Autonomous Agents (Agents ’98), New York 1998. ACM Press.
[28] A. Skvortsov, R. Connell, P. Dawson and R. Gallis, “Epidemic Spread Modelling:
Alignment of Agent-based Simulation with a SIR Mathematical Model,”
[29] Complex Warfighting Edition Two. ADF Future Land Operational Concept (FLOC) document.
[30] C. Heinze,Modelling Intention Recognition for Intelligent Agent Systems.PhD The- sis, Department of Computer Science and Software Engineering, University of Mel- bourne, Melbourne, Australia, 2003.
[31] C. Heinze, B. Hanlon, M. Turner, K. Bramley, J. Rigopoulos, D. Marlow and K.
Bieri, “The ARTEMIS Air-to-Air Combat Model,” in Proc. of SimTecT ’04, the Simulation Technology and Training Conference, Canberra Australia, 2004.
[32] D. McIlroy, B. Smith, C. Heinze, and M. Turner, “Air Defence Operational Analysis Using the SWARMM Model,” inProc. of Asia Pacific Operations Research Sympo- sium, 1997.
[33] J. Lee, M. J. Huber, P. G. Kenny, and E. H. Durfee, “UM-PRS: An Implementation of the Procedural Reasoning System for Multirobot Applications,” in Proc. of the Conference on Intelligent Robotics in Field, Factory, Service, and Space (CIRFFSS), Houston, Texas, 1994, pp. 842–849.
[34] M. d’Inverno, D. Kinny, M. Luck, and M. Wooldridge, “A formal specification of dmars,” inIntelligent Agents IV: Proceedings of the Fourth International Workshop on Agent Theories, Architectures and Languages, number 1365 in Lecture Notes on AI, Springer, 1998, pp. 155–176.
[35] G. Tidhar, C. Heinze, and M. Selvestrel, “Flying Together: Modelling Air Mission Teams,” inApplied Intelligence, vol. 8, pp. 195-218, 1998.
[36] M. Papasimeon,Intelligent Environments for Agents.PhD Thesis (in preparation), University of Melbourne, 2007.
[37] J. R. Boyd, “A discourse on winning and losing,” Technical report, Air University, Maxwell AFB, Alabama, USA, 1987.
[38] N. Howden, J. Curmi, C. Heinze, S. Goss, G. Murphy, “Operational Knowledge Representation: Behaviour Capture, Modelling and Verification,” in Proc. of the Eighth International Conference on Simulation Technology and Training (SimTecT
’03), Adelaide, Australia, May 2003.
Acknowledgment
The authors would like to thank the large number of air combat analysts, engi- neers and scientists who have contributed to the agent research and development programme over the last two decades.
Clint Heinze, Michael Papasimeon, Simon Goss, Martin Cross and Russell Connell 506 Lorimer St
Fishermans Bend, Victoria 3207 Australia
e-mail:[email protected] [email protected] [email protected]