Dignum F., et al. (Eds.). Cognitive Agents for Virtual Environments

Springer-Verlag Berlin Heidelberg, 2013. — 194 p. — ISBN 978-3-642-36443-3, e-ISBN 978-3-642-36444-0.First InternationalWorkshop, CAVE 2012, Held at AAMAS 2012, Valencia, Spain, June 4, 2012, Revised Selected Papers.There is a wide range of activity within the agent community considering various aspects of multi-agent systems, both theoretical as well as practical. This includes communication, team work, coordination, and cooperation of agents. In the First International Workshop on Cognitive Agents for Virtual Environments (CAVE-2012) we explored how these results might be used in the context of games and other virtual applications that require interaction with real users and perhaps identify any additional requirements that should be imposed for these contexts. We also explored similarities between solutions developed within the agent community with those used by people studying cognitive architectures.
The workshop brought individuals working on virtual characters together with those working on agent platforms and languages and cognitive architectures. All three communities have important parts of solutions for creating agents for games and similar applications, but very little is currently being done to combine these solutions. Thus the workshop hoped to connect the different communities and show the benefits from this combination. Although some cross-fertilization took place there is still room for improvement. However, the proceedings give a good indication of the state of the art in this area.
The workshop builds upon the AAMAS AGS 2009/10, EduMAS 2009, and AEGS 2011 workshops where the main issue has been to incorporate elements of agent technology in games and similar virtual environments such as 3D training and educational applications to create more flexible and realistic game play. Although some of the technical issues have been overcome and middleware (such as Pogamut, EIS, and CIGA ) has been developed to connect agent platforms to games like Unreal Tournament, there are a number of fundamental challenges both on the technical as well as on the conceptual and design level.
In these proceedings we include papers that address several of these challenges and were presented at CAVE 2012 held on June 4 in Valencia, Spain, in colocation with AAMAS 2012. We received 14 submissions of high quality covering many of the aspects mentioned above. Each submission was reviewed by at least three Program Committee members. We accepted 10 papers for presentation, which can be found in this proceedings. Afterwards we invited several persons to submit additional contributions for this volume in order to make the overview complete. However, we only selected one high-quality contribution of these submissions (reflecting that quality is more important than completeness).
We have grouped the papers into four sections. The first section contains papers that are related to architectures combining agents and game engines.
The paper of Tomas Plch, Tomas Jedlicka, and Cyril Brom discusses the use of HLA, a standard for coupling simulations, for coupling agents to game engines. HLA seems to provide a useful language to define (part of) such a coupling. The second paper by Jeehang Lee, Vincent Baines, and Julian Padget also looks at the more practical side of the coupling and discusses, for instance, performance issues related to tight and loose coupling. The last paper in this section from Joost van Oijen and Frank Dignum discusses the information flow between agents and game engines that is necessary to generate realistic communication between virtual characters. In the second section we included papers that focus on using agents and virtual environments for training team work. It is an excellent area where the use of cognitive agents is imperative for good results, but also where many issues are still open. The paper of Martin Beer, Emma Norling, Peter Wallis and Lyuba Alboul discuss how agents and controllers can work together in order to get (unmanned) aircrafts to discover lost persons in mountainous areas as quickly and efficiently as possible. The second paper in this section is from Nader Hanna and Deborah Richards and looks at the very essential issue of two-way human–agent communication within the context of an educational game. One of the prime challenges is the combination of verbal and non-verbal communication in this context. The paper from Marie Manner and Maria Gini relates how agents can assist people to improve team performance. The third section contains papers that describe how cognitive agents can be used for simulation environments. Both visualization issues as well as issues of efficiency and scale play an important role in this area. The visualization issue is the focus of the first paper in this section, which is from Athanasia Louloudi and Franziska KlЁugl. The second paper, from Quentin Reynaud, Etienne de Sevin, Jean-Yves Donnart, and Vincent Corruble, breaches the topic of combining cognitive and reactive architectures for urban simulations. The last section groups some papers around performance issues of cognitive agents for virtual environments. The paper from Surangika Ranathunga and Stephen Cranefield discusses the problem of interpreting useful events from the low-level data that an agent receives from the game engine. Having an efficient translation of this low-level information is important for the cognitive agents being able to react to it on time. The second paper in this section is from Rudolf Kadlec, Michal Cermak, Zdenek Behan, and Cyril Brom and discusses issues in generating corpora of daily living memories for cognitive agents. These agents should have some history to be believable, but this history should of course also be manageable! The last paper in this section discusses the tools that should be used to define agents for games. This paper from Jakub Gemrot, Zdenek Hlavka, and Cyril Brom describes an experiment in which different agent behavior specification tools are compared and the authors check whether high-level specification of behavior leads to higher productivity. Read the paper to check the results!
All in all we are very happy with the papers contained in this volume. We are sure they form a valuable overview of the current state of the art of cognitive, intelligent characters in virtual environments. Finally, we would like to thank the Program Committee members and external reviewers without whom the reviewing would not have been possible and who gave valuable comments on all papers.