MultiAgent GeoSimulation is a great technology which is used to populate the dead virtual geographical environment with entities that are intelligent, autonomous and interactive. The technology is extremely valuable to predict how people interact with the geographical environment and assess environments which do not exist but will be constructed in the future
Over the past decade or so, there has been increasing interest in simulation applications of human behaviors, with special focus on pedestrian behavior in spatial environments. Simulating urban phenomena or pedestrian/crowd movements has attracted the attention of both urban planners and government officials alike. Besides, it has also attracted retailers, advertising agents, and people involved in urban management. Some of the recent generation of models deployed a multi-agent system (MAS) based approach in order to represent the movements of people in virtual geographic environments. In most cases, these virtual geographic environments are obtained from a GIS. Combining MAS and GIS for simulation purposes gives rise to a new simulation approach called MultiAgent GeoSimulation.
Since the invention of 3D GIS and various date collection technologies such as 3D lasers, LiDAR, etc, several companies as well as government user organisations from different countries around the world have tried to develop virtual 3D geographic environments which reflect real countries or cities. A reason for doing that is because our world is 3D and so the virtual world should also be 3D in order to show us the true picture. Unfortunately, when we look at the 3D virtual environments, whether they represent cities or countries, they are empty like ghost world or cities (see Fig.1). On the other hand, our world is alive and full of visible and invisible entities. Moreover these entities, such as humans, animals, plants, etc. are intelligent and autonomous. In such a situation, geospatial technology and 3D GIS, in particular, represent incomplete reality.
 Fig 1: 3D virtual geographic environment of Dubai city |
In order to complete the representation of the real world (full 3D, alive and intelligent environment), we should populate our virtual environment with intelligent virtual entities. MultiAgent GeoSimulation, is a technology which is used to add intelligent entities to a 2D or 3D geographical virtual environment. MultiAgent GeoSimulation (MAGS) is powerful technology which combines: (1) MultiAgent technology, which is a field of distributed artificial Intelligence and (2) GeoSimulation, which simulates systems using GIS.
MultiAgent GeoSimulation is a great technology which is used to populate the dead virtual geographical environment with entities that are intelligent, autonomous and interactive. This technology is different from computer animation in the sense that in computer animation the entities are not intelligent but animated/controlled by the user or via specific patterns of behavior. In the following figures, we have presented two simulations. The first one represents the simulation of crowds of people in front of Quebec Parliament, in Quebec, Canada (See Fig. 3) and the second one represents the shopping behavior at the Square One Mall in Toronto (See Fig. 4).
 Fig. 3: The simulation of human behavior in front of the Quebec Parliament, Quebec City, Canada |
 Fig. 4: The simulation of human shopping behavior in the Square One Shopping Mall, Toronto, Ontario, Canada |
MultiAgent GeoSimulation is also used to assess or predict how people interact with the geographical environment. It can be used to assess environments which do not exist but will be constructed in the future. For example, we have changed the configuration of the virtual environment of Square One mall in order to see how the shoppers behave and interact with it. We have simulated two scenarios with two configurations and compared them. After the comparison, we can select the best scenario and apply it in the real world.
 Fig. 5: Comparison of two scenarios in the simulation |
Apart from that MultiAgent GeoSimulation can also be used to simulate panic behavior in a virtual geographical environment. This technology can be used to simulate and assess panic behavior and how people can interact with the environment in a crisis or dangerous situation such as fire or any other attack. For example, we have simulated a fire in the virtual shopping mall (Square One) and we have seen how virtual people behave in order to assess if the mall has enough fire exits or not.
MultiAgent GeoSimulation is used to simulate any intelligent behavior in geographical environment such as the passengers’ travel behavior in an airport in order to assess the capacity of a particular airport.
 Fig. 7: MultiAgent GeoSimulation of the travel behavior in an airport |
Conclusion
The above examples show how MultiAgent GeoSimulation technique can be used to simulate ‘knowledge-based’ human behaviors in micro-scale geographic environment. The article focuses on the agents’ individual features and on their behavior related to the apprehension of space (perception and memorisation). For example, the technique was implemented to create an application which simulates customers’ shopping behaviors in a mall. The results of the research show that the MAGS technique has great potential to simulate ‘knowledge-based’ human behavior in micro-scale geographic environments. Indeed, this technique can be easily adapted to simulate other kinds of behaviors in different micro-scale geographic environments.
