Coordinating Robotic Tasks and Systems with rFSM Statecharts

Abstract

Coordination is a system-level concern defining execution and interaction semantics of functional computations. Separating coordination from functional computations is a key principle for building complex, robust and reusable robotic systems. This work introduces a minimal variant of Harel statecharts called rFSM designed to model coordination of robotic tasks and systems with a minimal number of semantic primitives. Firstly, the semantics of the rFSM language are derived by analyzing state-of-the-art discrete event models and implementations and extracting a motivated and semantically well-defined subset that is considered best practice for the domain of robotic coordination. Secondly, a real-time capable reference implementation of rFSM is presented, which has been loosely integrated into the OROCOS/RTT framework. The application of rFSM is illustrated using a detailed description of a dual robot coordination problem. Lastly, several best practices and patterns are presented with the goal of i) supporting development of robust Coordination models, ii) illustrating how limitations of the statechart model can be overcome by extending the execution semantics, and iii) offering guidance in designing pure coordination components that optimize reusability.