AI Modelling: Behaviour Trees

Student thesis: Master Thesis and HD Thesis

  • Anders Tankred Holm
  • Mads B√łgeskov
2. term, Computer Science, Master (Master Programme)
The report documents the development of a set of tools and techniques, which goal is to aid the process of defining probability values in a behaviour tree. A formal model of behaviour trees is presented that forms the basis for the problem. The report presents measures for diversity and agent outcome for behaviour trees. An optimal selector probability distribution guideline is achieved by utilising linear programming using these measures, that then afterwards is used in the tool extensions. The project is then evaluated by a series of test subjects that concludes the presented techniques is beneficial.
Publication date2 Jun 2010
Number of pages108
Publishing institutionAalborg Universitet


ID: 32308291