Projects like Rome Reborn [14,15] have met enormous success, but what immediately becomes noticeable is that
those projects are ‘empty’ — no inhabitants crowd the streets of Imperial Rome. However, it is argued that
historians should also capture a degree of understanding on how the inhabitants of the heritage site would have
interacted with each other. This has been done in projects like the reconstruction of the City of Uruk [16,17], where
intelligent agents follow certain rules to depict common activities of local inhabitants, and allow human visitors to
interact with them to ask them what they are doing. Not only does this allow historians a better understanding of why
the architecture looks like it does, but also because it provides human visitors with a higher degree of educational
value.
Morgan [13], however, warns that non-playing characters (NPCs) should not be used as merely decorative
elements or ‘props’. In her critical analysis, she discovered that there was often the temptation of limiting NPCs to
chatting about their own environment, but be presented decontextualized from the heritage site; this, she argues, not
only lowers the credibility of the project, but ‘[...] turning people of the past into mere mouthpieces for their
architecture diminishes the rich potential of reconstructions to impart information about complex lifeways’. The
Uruk project avoids this trap, and so do the reconstructions done by Kennedy et al. [18]. In the Crystal Palace
project, NPCs which chat with each other and with visitors are employed to provide an understanding of what
Victorian visitors of the Crystal Palace exhibit of Pompeii thought about it, based on actual records of interviews
done at the time [19]. Thus, NPCs, if employed at all, have to be meaningful for the project.
The problem here, of course, is how to programme them, because historians and archaeologists might not have
the necessary skills. The above-mentioned projects all require specialised teams of programmers to do the complex
artificial intelligence (AI) programming, and leave the validation details to the historians.
In this article, a suggestion is proposed for a model where historians, instead of being passive towards the AI
simulation of NPC, become the ‘programmers’ themselves, but without requiring writing a single line of traditional
computer programming code. In other words, just like historians can move objects around in a virtual world (and
change environment settings) to formulate hypothesis and test theories, without requiring any assistance from the
technical teams (once the models have been built, historians can simply use them and place them on the required
spots), this article suggests that they should be given an easy-to-use tool to simulate crowd behaviour.
3. The City & Spectacle Project
City & Spectacle: A vision of pre-1755 earthquake Lisbon [20] is a project promoted by CHAIA/Universidade de
Évora with Beta Technologies as its computational development partner. It is a virtual archaeology project to
recreate a memory of a heritage site that does not exist any longer: the Baroque city of Lisbon, utterly destroyed by a
devastating earthquake, tsunami, and subsequent fire on November 1st, 1755.
Besides merely modelling architecture, the project also includes the simulation of specific events, namely, a night
at the Opera House, attending the coronation ceremony of King Joseph I, a bullfight at the Royal Court (Terreiro do
Paço) and following a processional across the streets of Lisbon — as well as the earthquake itself. Ideally, NPCs
would populate the streets of Lisbon, walking from home to work and back, and interacting with other NPCs they
would find on their way. Once one of the ‘special’ events is flagged as starting, some NPCs are expected to break
from their routine and attend the events.
4. Algorithms for AI-driven NPCs applied to Second Life/OpenSimulator
A discussion of the typical algorithms employed to appropriately model such a framework is beyond the scope of
this article. It is an area of active research for at least the past two decades [21–23], especially because of computer
games [24–27], which have similar specifications, but similar techniques are also employed in military simulations
[28,29]. Each approach makes certain assumptions about the environment in order to present a ‘best’ Artificial
Intelligence (AI) algorithm for developing autonomous agents. Early work by Wooldridge & Jennings [30] divided
the then existing solutions in two categories: planning and non-planning. In the first case, the scenario is deemed to
be reasonably static, and some sort of rule-based engine is used to determine in advance the ‘best’ solution for
movement/interaction within the environment. While individual agents will have exceptions to deal with sudden
changes, the simulation is expected to have a certain degree of stability. Non-planning methods, by contrast, do not
Projects like Rome Reborn [14,15] have met enormous success, but what immediately becomes noticeable is that
those projects are ‘empty’ — no inhabitants crowd the streets of Imperial Rome. However, it is argued that
historians should also capture a degree of understanding on how the inhabitants of the heritage site would have
interacted with each other. This has been done in projects like the reconstruction of the City of Uruk [16,17], where
intelligent agents follow certain rules to depict common activities of local inhabitants, and allow human visitors to
interact with them to ask them what they are doing. Not only does this allow historians a better understanding of why
the architecture looks like it does, but also because it provides human visitors with a higher degree of educational
value.
Morgan [13], however, warns that non-playing characters (NPCs) should not be used as merely decorative
elements or ‘props’. In her critical analysis, she discovered that there was often the temptation of limiting NPCs to
chatting about their own environment, but be presented decontextualized from the heritage site; this, she argues, not
only lowers the credibility of the project, but ‘[...] turning people of the past into mere mouthpieces for their
architecture diminishes the rich potential of reconstructions to impart information about complex lifeways’. The
Uruk project avoids this trap, and so do the reconstructions done by Kennedy et al. [18]. In the Crystal Palace
project, NPCs which chat with each other and with visitors are employed to provide an understanding of what
Victorian visitors of the Crystal Palace exhibit of Pompeii thought about it, based on actual records of interviews
done at the time [19]. Thus, NPCs, if employed at all, have to be meaningful for the project.
The problem here, of course, is how to programme them, because historians and archaeologists might not have
the necessary skills. The above-mentioned projects all require specialised teams of programmers to do the complex
artificial intelligence (AI) programming, and leave the validation details to the historians.
In this article, a suggestion is proposed for a model where historians, instead of being passive towards the AI
simulation of NPC, become the ‘programmers’ themselves, but without requiring writing a single line of traditional
computer programming code. In other words, just like historians can move objects around in a virtual world (and
change environment settings) to formulate hypothesis and test theories, without requiring any assistance from the
technical teams (once the models have been built, historians can simply use them and place them on the required
spots), this article suggests that they should be given an easy-to-use tool to simulate crowd behaviour.
3. The City & Spectacle Project
City & Spectacle: A vision of pre-1755 earthquake Lisbon [20] is a project promoted by CHAIA/Universidade de
Évora with Beta Technologies as its computational development partner. It is a virtual archaeology project to
recreate a memory of a heritage site that does not exist any longer: the Baroque city of Lisbon, utterly destroyed by a
devastating earthquake, tsunami, and subsequent fire on November 1st, 1755.
Besides merely modelling architecture, the project also includes the simulation of specific events, namely, a night
at the Opera House, attending the coronation ceremony of King Joseph I, a bullfight at the Royal Court (Terreiro do
Paço) and following a processional across the streets of Lisbon — as well as the earthquake itself. Ideally, NPCs
would populate the streets of Lisbon, walking from home to work and back, and interacting with other NPCs they
would find on their way. Once one of the ‘special’ events is flagged as starting, some NPCs are expected to break
from their routine and attend the events.
4. Algorithms for AI-driven NPCs applied to Second Life/OpenSimulator
A discussion of the typical algorithms employed to appropriately model such a framework is beyond the scope of
this article. It is an area of active research for at least the past two decades [21–23], especially because of computer
games [24–27], which have similar specifications, but similar techniques are also employed in military simulations
[28,29]. Each approach makes certain assumptions about the environment in order to present a ‘best’ Artificial
Intelligence (AI) algorithm for developing autonomous agents. Early work by Wooldridge & Jennings [30] divided
the then existing solutions in two categories: planning and non-planning. In the first case, the scenario is deemed to
be reasonably static, and some sort of rule-based engine is used to determine in advance the ‘best’ solution for
movement/interaction within the environment. While individual agents will have exceptions to deal with sudden
changes, the simulation is expected to have a certain degree of stability. Non-planning methods, by contrast, do not
การแปล กรุณารอสักครู่..