Archive for March, 2006

conversation with Rick on crowd (1)

Thursday, March 9th, 2006

INTRO
Lunch yesterday (8 March 2006) had interesting moments. As well as catching up on news after my 6 month stint in Australia, I was able to discuss some of the observer and defining characteristics (eg prediction) that I had previously come across in my reading on emergence, with colleague and professor Rick McKenzie.

Rick is the principal investigator of the Crowd Modeling project.
http://www.vmasc.odu.edu/crowdwebsite/Crowd_Project.htm
Here groups of characters are created by modeling the behaviours of individuals, and crowd behaviours emerge. Simple rules on individual entities result in complex behaviours that are not predictable. If the same random seed is used, the same result is repeated. The emerging crowd behaviour is the aim.

Birchko simulation by the police station with Non-Lethal Weapons action Crowd simulation at the Target Building.jpg
the chase

CONDITIONS FOR EMERGENCE
Rick considers the simulation to be successful when an observer cannot distinguish wether the crowd has been modelled (1) directly or (2) indirectly, through modeling the individuals with the crowd emerging.

Other points discussed are repeatability: this is considered desirable but given things like packet corruption or order arrival change; over a simulation that uses data across the network; it is only existant on a larger scale, not from one particular sim to the next.