Module Interactions for Model-Driven Engineering of Complex Behaviour of Autonomous Robots

Abstract

In this paper, we describe a model-driven engineering approach that enables the complete description, validation, verification and deployment of behaviour to autonomous robots, directly, and automatically from the models. This realises the promises and benefits of model-driven engineering, such as platform-independent development and behaviour traceability. However, such a top-down approach of modelling by finite-state machines and sub-machines creates a conceptual challenge to the behaviour designer due to the complex interaction of independent modules. Simply finding which modules are necessary for other modules can be a challenge. We also describe here our solution to this. Interestingly, our approach goes in the opposite direction of Object Oriented Software Engineering as currently represented by the Unified Modeling Language and corresponding software processes. That is, typically, the static models are derived first (and in particular class diagrams), while dynamic modelling follows later with behaviour diagrams and interactions diagrams. We actually start with the description of behaviour in finite state machines and we complement this by static information provided by logics that describe concepts and by our dependencies