A Reference Implementation for Knowledge Assisted Robot Development for Planetary and Orbital Robotics
Mehmed Yüksel, Thomas M. Röhr, Marko Jankovic, Wiebke Brinkmann, Frank Kirchner
In Acta Astronautica, Elsevier Ltd., volume 2023, pages 1-16, 2023.
Zusammenfassung (Abstract)
:
The use of modular, yet reusable and interchangeable components offers flexibility for system designers and
thus leads to a higher degree of reconfigurability with cost effective, adaptable and scalable solutions. The
development of a robotic system remains, however, a complex task with increasing demands depending on the
targeted capabilities, functionalities and operational constraints of the robot. This requires expert knowledge
in mechanical, electrical and software engineering, as well as tools that can collect, process and use relevant
data. As a result, the amount of available and managed information is increasing, while its relevance, usability,
and knowledge that can be extracted from it are consequently decreasing: a situation described as information
overload paradox.
The design of a modular system can be based on the use of standardized modules with universal
interconnects, but it should be aligned with a strong formalism, e. g. the semantic representation of components,
to allow for a better management of the overall complexity. An ontology-based knowledge representation, for
instance, which provides domain and application specific knowledge for the design of robotic systems, can be
used as a method of storing and sharing data in a uniform, machine-readable and standardized way.
In this paper, we introduce the Knowledge-based Open Robot voCabulary as Utility Toolkit (korcut) as
a core component to support the ontology-driven development of robotic systems as part of a reference
implementation of the Q-Rock development cycle. Q-Rock gathers methods to assist users in designing robot
configuration by: (a) automatically exploring robot capabilities based on robot hardware, (b) proposing robot
designs to meet the user’s needs, (c) refining the proposed robot designs. We use korcut to improve the
robot design process in orbital and planetary robotics from the requirements definition to the development
phases of complex modular robotic structures (e. g. composition of an unmanned ground vehicle with robotic
arms). This is implemented by embedding semantic component descriptions in a state-of-the-art open-source
3D modeling software. Furthermore, the korcut ontology family includes various sub-ontologies to address
specific space-related tasks, such as the modeling of a Standard Interconnect (SI) for On Orbit Services (OOS)
and Orbital Factory. The ontology design follows criteria that have been derived from a survey conducted as
part of this work. This paper also covers an evaluation of its current applications from a methodological,
knowledge representation, and software tool perspectives. In the application part, we introduce the use of
the ontology to model a multi-functional SI that enables mechanical connections between various (modular)
robotic components, as well as transfer of power and data. Finally, we provide a critical analysis of our work
and outline its future work.
Stichworte
:
Robotics, System design, Domain ontology, Space interface benchmark, In-orbit servicing, Planetary robotics
Links:
https://www.sciencedirect.com/science/article/abs/pii/S0094576523002485