The Robot Architecture Framework - A systematic approach for describing the architecture of complex, autonomous robotic systems.

Abstract

Many autonomous robots are currently designed based on the system architect’s individual experience and knowledge, taking into account given factors such as application and environment. The complexity of robots, but also the diversity of systems, applications and fields of use, make it difficult to compare these individual solutions and to transfer architectural approaches to new system designs. This thesis answers the research question of how the knowledge about robot architecture contained in existing systems can be made accessible. The developed process enables the systematic capturing of the architecture of autonomous robot systems. The resulting structured architecture descriptions make it possible to identify similarities and differences in the architectures of different systems despite the complexity and heterogeneity of robots. The methodology chosen is an extension of the Architecture Framework concept. This approach, which originates from the field of software development, is used to capture the architectures of distributed software systems. This work transfers and extends the approach to the domain of robotics by defining a Robot Architecture Framework. The architecture description process based on this framework enables the architecture of existing robotic systems to be identified. The applicability of the process is demonstrated using three very different robotic systems. It is shown that compact architecture descriptions can be generated despite the complexity and heterogeneity of the systems. In addition, the broad applicability of the defined process is also conceptually validated.