Language and Code Generator Composition for Model-Driven Engineering of Robotics Component & Connector Systems

Abstract

Engineering software for robotics applications requires multi-domain solutions. Model-driven development (MDD) promises efficient means for developing domain-specific and reusable models of robotics software. Code generators transform these models into executable code for specific robotic platforms. Robotics is heterogeneous and robotics applications pose various challenges to leveraging the potential of MDD. Combinations of modeling languages and platforms are often problem-specific. Generative software development for multi-domain applications requires both the effective integration of modeling languages and composition of code generators. We present the extensible MontiArcAutomaton architecture modeling framework for the generative development of robotics applications with a strong focus on language integration and generator composition. Its core modeling language is a component & connector architecture description language that can be extended with problem-specific component behavior modeling languages. We present how MontiArcAutomaton supports syntactic and semantic behavior modeling language integration to describe component behavior in most suitable modeling languages and code generator composition to synthesize code from integrated models. We sketch a process for model-driven development of robotics applications using MontiArcAutomaton.