Catalogue of Artificial Intelligence Techniques
Collective Locomotion (Egress Alogorithm)
Aliases: Egress Algorithm
Keywords: collective robotics, locomotion, movement, navigation
Author(s): John Fraser
Although research into collective locomotion is still ongoing, there have already been successful projects, using implementations of the Egress Algorithm. This algorithm is based on a while loop that is passed a location to get to, and during the attempts to reach said location sets its own sub-goals based on the surroundings. Having classified its surroundings into a pre-constructed category, it can then choose and apply the appropriate rule, and as such set the agent a new sub-destination point and move there, checking for possible collisions and updating the agent's location. As such this is a form of self optimization, as the algorithm is continuously updating the best possible sub-route/s to be taken in order to reach the end objective, taking into account other agents in proximity. As a result, the Egress Algorithm has replicated human crowd behaviour in various implementations with fair success, as crowd phenomena such as panic can be replicated by altering the velocity and collision rules for the agents. By doing so, agents become more aggressive in their route planning, and less inclined to negotiate a mutually beneficial route.
- Xiaoshan Pan, Charles S. Han, Ken Dauber and Kincho H. Law, Human and social behavior in computational modeling and analysis of egress Automation in Construction 15 (2006) no.4, July.