Recent technological advances have begun to make robots more affordable and, as a result, more common not only in industry but also in the home. A recent report by the United Nations Commission for Europe (UNECE) has indicated that robot orders in the first half of 2003 increased by 26% and worldwide growth during 2003-2006 is expected to reach an average of 7.4% per annum. Furthermore, "[a]t the end of 2002 more than 50,000 autonomous vacuum cleaners and lawn-mowing robots were in operation. By the end of 2006, a tenfold increase is predicted."
While traditional robotics has focussed on low-level aspects such as localisation, navigation, sensor fusion, etc., very little attention has been paid to more high-level issues. In particular, very little work has been done on the higher level, cognitive aspects of robot control and the development of architectures that allow one to easily integrate the diverse software modules that may be put together to create a complex robotics application.
The aim of this project is to develop a robust architecture that allows one to integrate diverse software modules (e.g., speech, natural language, low-level controllers and high-level, deliberative controllers) into a sophisticated robotic control system. The architecture should not depend on the robotic hardware platform on which the system is to run but be easily portable across architectures. There is a large amount of work in this area on software agents (e.g., SRI's Open Agent Architecture and JADE) but these types of frameworks tend to be too large and cumbersome for robotics applications. Our work will focus on robotics applications in particular, providing a seamless mechanism for integrating diverse components in a rigorous, provably correct manner. An outcome of this research is an industrial standard to be used for the integration of software modules when developing robotics applications.