We have just published a systematic review article in the journal Frontiers in Robotics and AI. The article, titled A survey of ontology-enabled processes for dependable robot autonomy analyses the use of formal ontologies in the improvement of robot autonomy.
The article has been the product of the effort of Esther Aguado, as part of the work of her excellent PhD Thesis (with some important collaboration from the rest of the authors in the context of the Horizon 2020 project ROBOMINERS and the Horizon Europe project CORESENSE).
A summary of the article:
Autonomous robots are already present in a variety of domains performing complex tasks. Their deployment in open-ended environments offers endless possibilities. However, there are still risks due to unresolved issues in dependability and trust. Knowledge representation and reasoning provide tools for handling explicit information, endowing systems with a deeper understanding of the situations they face. This article explores the use of declarative knowledge for autonomous robots to represent and reason about their environment, their designs, and the complex missions they accomplish. This information can be exploited at runtime by the robots themselves to adapt their structure or re-plan their actions to finish their mission goals, even in the presence of unexpected events. The primary focus of this article is to provide an overview of popular and recent research that uses knowledge-based approaches to increase robot autonomy. Specifically, the ontologies surveyed are related to the selection and arrangement of actions, representing concepts such as autonomy, planning, or behavior. Additionally, they may be related to overcoming contingencies with concepts such as fault or adapt. A systematic exploration is carried out to analyze the use of ontologies in autonomous robots, with the objective of facilitating the development of complex missions. Special attention is dedicated to examining how ontologies are leveraged in real time to ensure the successful completion of missions while aligning with user and owner expectations. The motivation of this analysis is to examine the potential of knowledge-driven approaches as a means to improve flexibility, explainability, and efficacy in autonomous robotic systems.
Get the article:
And also Aguado’s Thesis: