Case Study: Darug Clan, Australia

In our second case study, we simulate the life of an Australian Aboriginal clan from the Darug tribe, living in the area of Parramatta, New South Wales, in times before the arrival of first fleet and the establishment of the European settlement.

The interactive 3D video game takes the player on a quest to explore the life of an Aboriginal clan in the Parramatta basin. A spiritual mentor and the guardian in the form of an aboriginal elder gradually introduces the participant to the daily life of native clans, the knowledge they possessed, rituals they performed, protocols they kept and their connection to dreamtime. The elder familiarises the player with vari­ous clan members as they perform their daily activities such as tool making, painting, fishing or preparing food. During these interactions the player also learns about the aboriginal medicine, arts, as well as ceremonies, such as the smoking ceremony and receives an introduction to their spiritual values.

The virtual reality simulation uses Oculus Rift (6) headset to take the user on an immersive journey in historical Australia. The information provided is the same as in the video game, yet the content is not interactive, and user partakes the role of a sole observer listening to the spiritual mentor. In the remainder of this section, we describe how we applied our methodology to deliver this historical simulation and discuss the believability of our approach.

14.5.1 Preparation: Designing the World

The initial step of the 3D simulation creation process was the artistic design of 3D assets that formed the simulation environment. This phase highly depended on the invaluable help from the Elders of the Darug clan, who consulted us on the believ­ability of our simulation.

⁶https://www.oculus.com/ (last visited (06/2015)).

Fig. 14.17 Environment design. a Initial model (sketchup). b Final model (Unity)

with grounds, upon which the South Parramatta campus of the University of West­ern Sydney is located. The initial model of the environment was constructed from the GIS data using Google Sketchup^[13] (see Fig. 14.17a). We have modified this ini­tial model in Unity 3D and added local flora and fauna (see Fig. 14.17b). Then, we have implemented the animal behaviour (artificial intelligence), including the flock­ing behaviour of kangaroos, moving lizards and snakes, flying birds and bats and fish swimming in the river. While bats and lizards follow pre-defined paths, kanga­roos show intelligence by fleeing from moving humans and placing bait in the water attracts fish. We implemented the animal behaviour using popular Unity 3D plugins: Playmaker^[14] and NodeCanvas.^[15] Use of these plugins facilitates the reusability of the developed functionality, and its visual nature helps the non-technical team members actively participate in its specification.

With the static 3D design of the environment in place, we apply our methodology to populate the simulation environment with autonomous agents.

14.5.2 Step 1: Design the Base Population

The population in this simulation consists of a single ethnic group. Therefore, we designed only two members of the base population to generate the rest. While both designed avatars carry typical aboriginal features, such as the wide nose or dark hair, we have also introduced variations of visual properties, such as skin colours or height. Maintaining consistent values (or just a minimal variation) of stereotypical features in all designed avatars, significantly increases the probability of their prop­agation to descendants, and their exclusion can only be affected by the mutation.

Consequently, consulting aboriginal elders, historic literatures and photographies, we have designed traditional clothing, tools and wearables. Since aboriginal popu­lation in the simulated era wore clothing that often did not cover the private parts, in order to target wider audience in which some may be offended by nudity, we decided to use the design of more recent clothes. Figure 14.19 depicts the generated crowd

of males and females, wearing traditional hip or neck bags and variations of loin clothes from the kangaroo or possum fur. To assign random clothes to avatars, we have used the popular Unity Multi-Purpose Avatar (UMA) plugin.^[16]

14.5.3 Step 2: Configure Motivational Modifiers and Step 3: Specify Personality Traits

Physiological motivation drives agents’ proactive goal creation. Motivational mod­ifiers affect the pace in which they get hungry, thirsty or tired. Having varied values assures that agents take decisions at various intervals. Motivational modifiers and personality traits form the part of the chromosome that is used by the genetic repro­duction. Since we used only two members of the base population, similarly to visual features, to gain variation in descendants, we have assigned motivational modifier values from both sides of feasible extremes. We created a male that was getting hun­gry at a high pace, needs to drink often, yet takes a long time to get tired. Female is humble on resources, yet fatigues quickly.

Concerning avatars personalities, we used the very same approach, providing sig­nificantly different OCEAN profiles for both avatars, with aggressive, yet timid male and submissive, yet social female.

14.5.4 Step 4: Formalise Social Norms and Roles

In this step, we defined components of the Electronic Institution. The dialogic frame­work contains roles of fisherman, spear-maker, hunter, tool-maker and gatherer (see Fig. 14.3b). All of these roles are sub-roles of person, which holds common proper­ties for all roles, e.g. inventory of owned items.

Please note that this role hierarchy does not correspond to the actual tribe struc­ture. In the real world, tribesmen perform various tasks, where individuals design their spears, create own tools and go hunting with them. This knowledge is incre­mentally passed onto by elders. We have separated the activities between various roles to isolate effectively agent behaviour and goals. Also, currently there exist only limited means for the dynamic plan creation with a large decision space. By dynamic plan creation we mean a possibility of agents to automatically generate and adapt plans that lead to the fulfilment of their goal (e.g. they do not need to re-start the whole plan, but continue from the last feasible point).

During the design of the Electronic Institution, we were able to re-use a large portion of scene protocols from the Uruk project (e.g. eating, gathering, fishing), supporting the usability of our approach. The reason we can reuse EI functionality is that protocols for action execution remained the same, what changed is their visu­alisation in the 3D environment. For example, in the fishing scene, one fisherman controls the boat while the other is using a spear to catch fish. The process is the same in Uruk as well as with Darug clan, but Darug fishermen no longer sit and stand in the boat, they kneel. Also, not a long paddle but a piece of bark is used to steer and row the boat. As a result, we only needed to record new animations for action execution, their control remained the same.

14.5.5 Step 5: Adaptation and Annotation of the

Environment

With the invaluable help from Aboriginal elders who performed as actors, we recorded authentic animations that portray daily chores and activities of the Darug tribesmen.

Fig. 14.20 Simulated activities. a Feasting. b Tool making

Fig. 14.21 Everyday life in the Darug clan

14.5.6 Step 6: Generating the Population

In the last step of our methodology, we generate the population of agents. Using our approach, the appearance, physiological and personality profile are unique with each generated agent.

For each created agent, a Playmaker script assigns a role and initialises the con­nection with the Electronic Institution. Agents start the random walk to discover their environment. After an arbitrary period, physiology triggers a feeding request. To feed, agents dynamically create plans based on their role, personality and sur­rounding objects. For example, the only way for spear-makers to feed is to exchange their produced spears for food with fishermen or hunters. In order to create a spear, they have to use knives, which serve them to work the spear-wood. Another role, Painters, first search for ochre, water and bark and then exchange their pieces with hunters and fishermen. Figure 14.21 shows a scene from the aboriginal simulation showing tool maker and painter at work, as well as men feasting on a kangaroo.

14.6