DesignLab
  • Departamento de Engenharia Mecatrônica
    Escola Politécnica - Universidade de São Paulo

GORE METHODS TO MODEL REAL WORLD PROBLEM DOMAINS IN AUTOMATED PLANNING

Javier Martinez Silva e José Reinaldo Silva

Journal 
 
Autor 
Javier Martinez Silva e José Reinaldo Silva 
Book Title 
Anais do Simposio Brasileiro de Automação Inteligente 
Volume 
Página(s) 
 
DOI 
 
Published 
2017 
Month 
Tipo de Documento 
Conference Proceedings 
Abstract 
In the field of intelligent design, the early phase, dedicated to requirements modeling and analysis plays a fundamental role, especially when analytic formal solutions are not suitable. Automated planning falls in that category - particularly when the target are ”real world” systems. In requirement analysis Knowledge Engineering is explored to provide clues that can facilitate a convergence for a good planning solution. Therefore, a great effort has been made today in the area of Artificial Intelligence to define a reliable design process for automated planning that includes Knowledge Engineering treatment in the early phase, coupled to requirements modeling and analysis. This paper presents an integrated approach to requirements analysis based on GORE (Goal Oriented Requirements Engineering) that starts by modeling a knowledge architecture based on a domain and planning requirements represented in KAOS and converted in Petri Nets (PN) to analysis. A software tool called rekPlan is proposed to generate the PN graph. The analysis is made in another software tool proposed in our Lab that, GHENESys (General Hierarchical Enhanced Net System), that support unified Petri Nets following ISO/IEC 15.909. A real case study is presented, based on classic problems of pre-salt petroleum industries. 
Keywords 
requirements engineering, GORE, Petri Nets 
Publisher 
SBAI 
URL