Publikation
A Holistic Approach for Enhancing Data Integrity and Reliability in Human-Robot Interaction
Manuel Meder; Kashmira Shinde; Dennis Hemker; Sadique Siddiqui; Teena Hassan; Nina Hoyer; Elsa Andrea Kirchner
In: Proceedings of SpaceCHI 2.0: Human-Computer Interaction for Space Exploration. SpaceCHI 2.0: Human-Computer Interaction for Space Exploration (SpaceCHI 2.0-2022), A Workshop at ACM CHI 2022, May 1, New Orleans, LA, USA, ACM, 5/2022.
Zusammenfassung
Space is a remote and hostile environment, and it poses several
challenges to both, humans and robots deployed in space missions.
This necessitates the use of robotic systems on which humans can
rely on. However, in extreme environments, sensors might fail or deliver
incorrect measurements. Furthermore, signal noise and model
inaccuracies could result in inconsistencies between information
extracted using different sensors or algorithms. Erroneous and unreliable
data can cause autonomous systems to behave in unexpected
or inappropriate ways, which reduces human trust in such systems.
This in turn would hamper the assistive potential of robots in space
missions. Methods that identify and handle sensor failures and lowlevel
data inconsistencies are commonly applied to strengthen the
system’s robustness. However, few have addressed logical inconsistencies
that occur at higher semantic levels of data processing.
By following a holistic approach that checks for, and handles both
types of inconsistencies, we can enhance the integrity of the data
that is used for higher-level inferences and decision-making by
robots. This would make robots a more reliable interaction partner
for humans. Therefore, we propose an architecture to detect and
respond to inconsistencies at both the local and the global levels of
data processing. A first implementation of the proposed approach includes auxiliary methods to perform preliminary checks and to
facilitate integration into a robot’s dataflow architecture. The proposed
architecture could also be applied to terrestrial use cases
(e.g., in healthcare), where data integrity, reliability, and trust play
a crucial role.