Introduction

Intelligent virtual¹ agents are autonomous computer programs that are represented in² a virtual reality environment by human-like (or animal-like) 3-dimensional figures (called avatars) that move around the reconstructed environment and simu­late its inhabitants.

With modern advancement in research and development, we are now reaching the stage when reconstructing a heritage site becomes more affordable. One possible way of reducing the development cost is to automate the design of the reconstructed virtual environment. Such design automation can be achieved with the use of design grammars—a procedural approach to generating historically informed designs of high complexity, allowing for large cities to be created in a matter of days rather than months. One of the well known examples of using this approach in histori­cal reconstructions is the Rome Reborn project (Dylla et al. 2009), where a virtual reconstruction of the entire city of ancient Rome in the period of 320 AD was gen­erated by automatically placing procedurally generated buildings onto a map of the

*See the prototype video at: http://youtu.be/ZY_04YY4YRo.

²See the prototype video at: https://www.youtube.com/watch?v=-jDsyOLZHN4.

T. Trescak (S) · A. Bogdanovych · S. Simoff

University of Western Sydney, Penrith, Australia

e-mail: t.trescak@uws.edu.au

A. Bogdanovych

e-mail: a.bogdanovych@uws.edu.au

S. Simoff

e-mail: s.simoff@uws.edu.au

© Springer International Publishing Switzerland 2016 377

J.A. Barcelo and F. Del Castillo (eds.), Simulating Prehistoric and Ancient Worlds, Computational Social Sciences, DOI 10.1007/978-3-319-31481-5_14 city produced by archaeologists.

Modern video games are a good illustration in regards to the possibilities that arise with employment of virtual agents in simulating human behaviour. Players of mod­ern games often experience complex human-like interactions with virtual agents and agents themselves become one of the most important aspects of game play and one of the key entertainment factors. Due to problems with automating agent development the cost of developing video games is very high and it is often hard to justify such high spending in non-profit areas of research that usually require cultural and his­torical simulations. An example that illustrates the magnitude of spending in video games is Crysis 3, a popular game with the estimated cost of developing being in excess of $66 Million (Gauder 2013). It’s hard to imagine such level of spending when it comes to historical simulations, so populating a historical environment with virtual agents needs to be automated.

Aiming to achieve cost saving, some researchers model their virtual societies at the level of crowds rather than individual agents. One well known example of using “virtual crowds” in historical simulations is the visualisation of the Roman Colos­seum (Gutierrez et al. 2007), where a crowd simulation approach is taken to visualise the spectators in a gladiator fight. While virtual crowds essentially consist of a large number of virtual agents, designing a crowd normally comes down to designing a few individuals and then replicating them a desired number of times with slight mod­ifications so that the crowd appears to be diverse. The state of the art in using agent crowds in historical simulations is outlined in Mam et al.

A number of crowd simulation and crowd generation approaches appear in the literature but hardly any of them advance beyond having avatars moving around and carrying objects with them. Further in the paper we show how through simulation of physiological needs and motivations together with personality traits we can achieve much more sophisticated simulations of human behaviour. Furthermore, employing genetic methods for inheriting personality traits and appearance characteristics and connecting virtual agents with formalisations of social roles and social norms allows for a similar (or event higher) level of complexity in large agent crowds as seen in commercial video games. In contrast to standard video games, however, the cost of development for such agents can be greatly reduced through a high degree of automation that our approach offers.

The remainder of the paper is structured as follows. Section 14.2 presents motiva­tion for selecting the combination of genetics, social norms, personality and physiol­ogy as a way of advancing the state of the art in historical simulations. Section 14.3 presents our methodology to be employed for creating such simulations. Section 14.4 shows how the aforementioned methodology was applied to building a historical sim­ulation of everyday life in the ancient city of Uruk, 3000 B.C. Section 14.5 shows the application of our methodology to an Australian cultural heritage case study, in which we simulate the life of an Aboriginal Darug tribe. In Sect. 14.6 we analyse the results obtained from the case studies. Finally, Sect. 14.7 summarises the contribu­tion and outlines the directions of future work.

14.2