Modelling Navigation and Landmarking in Ants
(Extends Model by) Kok Cheng Tan
Tkeden model of visual navigation in ants.
This project represents an attempt to model some of the various ways ants navigate, particularly their visual navigation and use of landmarks. As the project report explains, Empirical Modelling ideas were used to try and develop a lifelike pattern of ant behaviour.
The model itself shows a top-down view of the ant and the landscape, as well as a representation of the senses available to the ant. The landscape can be modified by the user to contain different coloured objects (landmarks). As the ant moves through the landscape, the ant's sensory information changes appropriately to reflect the surroundings.
The AI included in the model uses these modelled senses to navigate between the nest and the food. This involves a form of pattern matching between what the ant can currently see and 'snapshots' of desirable locations that the ant has recorded on its previous travels.
Though the AI developed in the project is far from comprehensive, it can produce some interesting and complex behaviours. Some of these behaviours can be quickly 'demo-ed' using saved scenarios in the model; the 'quick guide to using the model' included with the code explains how to do this.
This edition of Daniel Keer's model was prepared by Meurig Beynon for a Pupil Conference held in Computer Science on July 8th 2010. The minor changes to the 2005 version include fixing what seems to be a bug in the hidessmarkers() procedure, and customising the interface so as to support basic activities based on the model that are described in the associated Tour. This involved modifying the macro.e and macro2.e files, and introducing an additional file currsnapshot.s that can be used to assist manual execution of the route finding process.