4.01.2012

Part 3: Agent Based Modelling - Turning an Image into space





The next assignment in the Visualization course of the MRes Advanced Spatial Analysis and Visualisation in CASA, was relevant to "Agent based modelling". In this project the idea is to create a tool which loads the pixels of any image, recognise their contrast in comparison to their neighbours and finally produce points which interact with wandering Agents. In this case, the Agents are just moving points which perceive pixels as boundaries and change direction when they get close to them. 

However in this project, the Agents are not aware of each other. They move independently, creating random paths. "In  Multi-Agent Systems (MAS) the agents are autonomous entities that have certain characteristics and either cooperate or act selfishly. "[1] Even though the strategy of each individual entity is straightforward, these systems can aquire  complex behaviours and therefore can be used for problem solving. "Analysed agents are intended to accurately mimic real world entities, based on empirical data or ad hoc values that are realistic substitutes of observed processes."[2]

In that context, the next step is for these Agents is to become aware of each other and start having swarm behaviours. 

[1]  Definition of Multi-Agent Systems (MAS): http://www.dortmundintelligence.com/technology_ci_mas.html
[2] Batty Michael, [2005], "Cities and Complexity, Understanding Cities with Cellular Automata, Agent-Based Models, and Fractals", N. York: The MIT Press, p. 385

Assignment in the context of the visualization course by Martin Austwickin the mres ASAV 2011. 
music: Cannonball Adderley feat. Miles Davis " Autumn Leaves" (1958)

1 comment:

  1. Love this. Love the music. Love the flashing of the walls as agents bump into them. Beautifully understated.

    ReplyDelete