A Generalized Stochastic Petri Net model of route learning for emergency egress situations. (June 2018)

• Record Type: Journal Article
• Authors: Danial, S.N.; Khan, F.; Veitch, B.
• Is Part Of: Engineering applications of artificial intelligence. Volume 72(2018)
• Page Start: 170
• Page End: 182
• Publication Date: 2018-06
• Subjects: Route learning -- Generalized Stochastic Petri Nets -- Forgetting -- Remembering -- Landmarks -- Safe evacuation; Engineering -- Data processing -- Periodicals; Artificial intelligence -- Periodicals; Expert systems (Computer science) -- Periodicals; Ingénierie -- Informatique -- Périodiques; Intelligence artificielle -- Périodiques; Systèmes experts (Informatique) -- Périodiques; Artificial intelligence; Engineering -- Data processing; Expert systems (Computer science); Periodicals; 620.00285
• DOI: 10.1016/j.engappai.2018.03.024
• Languages: English
• ISSNs: 0952-1976

Abstract

Abstract: Route learning is an essential activity for a person visiting a new environment. The element of forgetting a location (called decision point) along a route, where a change in direction is needed, is of immense importance especially during emergency evacuation scenarios. It is this element that has not been given the attention it deserves in developing a route learning algorithm. This work proposes a model of route learning in a new environment based on landmarks using generalized stochastic Petri nets because landmarks based route learning has been observed as a method natural to humans. The model takes information about landmarks along a route and associated navigation commands and then chooses whether to save this information as part of the learned route or not. The selection is made by exploiting stochastic transitions for which the firing rates are dependent on the type of landmark encountered at a decision point. The final output is a route having some decision points missing; resembling the situation that humans encounter after they visit a route in a new environment. The model results closely match empirical results obtained with human subjects.