Friday, May 22, 2009

Agent Analyst movies

Agent Analyst is an Agent-based modelling extension that allows users to create, edit, and run RepastJ and RepastPy models from within ArcGIS. The Agent Analyst toolkit was developed by Argonne National Laboratory’s Center for Complex Adaptive Agent Systems Simulation in collaboration with ESRI and to promote cooperation and collaboration between GIS professionals and agent-based modellers

While its been around for some time, I have recently just come across two videos from the Redlands Institute, however the actual models and further details is limited, they show the potential of the extension for agent-based modelling. The first demonstrates the integration of Agent Analyst and Tracking Analyst within ArcGIS Desktop. Specifically the model shows a simulation of bird migration patterns of 2 species. The second model uses Agent Analyst to compare two urban growth scenarios.

Bird migration patterns

Comparing two urban growth scenarios

For more information including tutorials on how to use Agent-Analyst see the Agent Analyst website.

meta beta's Flu Model

Miles Parker writer of the meta beta blog has had some interesting posts recently about modeling influenza using agent-based models. In his first post he explained the design of simple agent-based model of influenza (including agents attributes and behaviours), in a second post he details some experiments (including the importance of carrying out batch runs) and the third post allows users to interact with the model through a applet. The model itself was developed using metaABM.

The effect of movement on the maximum infection Level (21 runs) (Source meta beta)

Miles is also the founder of Metascape and metaABM. metaABM is an opensource project which provides tools to allow for models to be visually edited without reqiring programming experience. Along with being tightly integrated within the Eclipse IDE toolset and can be linked to Repast Simphony. For more details readers are referred to the metascape website which has more information about the metaABM project along with some interesting videos on the overview of metaABM how one can install and design agent-based models in a relatively visual environment.

metaABM from Metascape

On a related note I found the movie below of an interview of Josh Epstein, were he talks about agent-based modelling and the spread of smallpox.

Wednesday, May 20, 2009

NetLogo to 3D Max

I am a great fan of NetLogo for the development of agent-based models. At CASA we have been exploring how one can loosely couple NetLogo models with 3D Max. The reason being that 3D Max offers a means of visualization which is beyond NetLogo's for example, building 3D city scapes.

Ateen Patel a PhD student at CASA and author of the Crowd Simulation Blog has written a short tutorial (including the code) and several other posts on how to get information from a simple NetLogo Traffic Model and to visualize the model in 3D Max.

Such outputs can be seen in the movie below from Digital Urban:

The Crowd Simulation Blog is worth keeping an eye on for those interested in pedestrian modeling as Ateen's research and his blog explores new methods for rendering very large crowd scenes that contain the correct kinds of physics and dynamics that enable realistic simulations to be generated.

Tuesday, May 19, 2009

Update: Agent-based models in Second Life

We just finished another part of Agent-Street where agents and avatars can interact with each other. The idea here is to merge iconic and symbolic urban models in a multi-user real time environment.

The movie below shows Agent-Street and how one can download the models we have created by clicking on the model vending machines. By clicking on the vending machines, models can be downloaded and saved. Such vending machines are common features in Second Life and allow users to store objects and scripts in their personal inventory for later use such as rebuilding the models on their own Land or in free sand-boxes. We have been using the Mauve Land for this (SLURL).

In our previous models avatars could only visualise and initiate the models. For example, in the pedestrian evacuation model users could only observe how agents exited the building (in a similar way to professionally developed 3D pedestrian modelling software packages such as STEPS). In this new model, we extend the basic pedestrian evacuation model, so that agents not only consider their environments but also other avatars. In this sense we are not only incorporating iconic and symbolic modelling styles but also adding a further human dimension, moving towards an augmented reality (i.e. as if the agents and the humans were in the same crowd). The movie below shows how pedestrian paths change when avoiding an avatar. The movie shows a how an avatar stationed near the exit impacts on pedestrian egress as agents have to move around it.

Previous posts on our and others work on Second Life can be found by clicking here. We have also set up a website outlining the work in detail (click here) or alternatively if you have a Second Life account this SLurl will take you to the Land, once you teleport to the area, follow the red arrow (beam of red light) into the sky to find Agent-Street. If the models are running slow let us know as we having problems with fish entering our land from a neighbouring island.

Friday, May 15, 2009


I been meaning to highlight the SIMPOP models for a while, and after listening to Denise Pumain talk at the AAG annual meeting a few months ago entitled "Modelling spatial evolution: the example of urban systems." I thought it was about time.

The SIMPOP model is basically a multi-agent model of cities systems evolution. In the model, agents are geographical entities: towns and cities. "This choice refers to evolutionary urban theory : the hypothesis is that the dynamics of a town or a city depends on its ability to interact with other towns and cities, which in turn depends on its relative situation in the settlement system (in terms of hierarchical level, specialization, accessibility). The city is then seen as a coherent entity, whose potential of growth depends first on its external interactions (core rules of the model) and second on internal specificities (governance). The interurban interactions are the driving forces shaping the emergence and evolution of the system of cities. (SIMPOP website)."

The model has been developed since 1996 and has been used to model the growth of Europe from 1300 and the United States of America from 1650.

To find out more about the models visit the SIMPOP Project website which has an introduction to the generic model, application domains and a list of publications resulting from the work (however, some of the links are currently not working).

Thursday, May 14, 2009

Tracking Pedestrians

Tracking peoples movement is an extremely time consuming task to do by hand however such analysis informs our ability to realistically model peoples moment in pedestrian models.

Researchers at the Computer Vision Laboratory, ETH have been working on multi-object tracking which combines object detection and space-time trajectory estimation. The ability to automate such a process obviously has benefits for understanding pedestrian movement in cities

To see a video of the work click here.


B. Leibe, K. Schindler, N. Cornelis, L. Van Gool. (2008). Coupled Object Detection and Tracking from Static Cameras and Moving Vehicles. In IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 30, No. 10, pp. 1683-1698.

A. Ess, B. Leibe, K. Schindler, L. Van Gool. (2008). A Mobile Vision System for Robust Multi-Person Tracking. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR'08), Anchorage, USA, June 2008.

Digital Urban pedestrian tracking tags