Project Description

The new generation of robots needs to be flexible, easy to program, and facilitate interactions with people. This encompasses a large variety of techniques that are usually developed individually. However, the robotics community has not yet solved a fundamental problem: how to fill the gap between the high-level decision and the execution systems. CSIC (experts in planning and knowledge representation) has proposed an influential solution to this problem (Canal et al. ICRA 2018). In this project, we plan to work in this direction by joining efforts with NSTC (experts in robot execution and parameterisation) who will provide robot execution primitives and parameter integration required to design and evaluate any new technology.

In scenarios where robots interact with humans, robots’ ability to model and reason about what they know becomes fundamental to develop reliable and trustworthy robots. Additionally, robots will need to understand how unexpected situations might trigger adaptation of their plans. In this project we will represent and reason about the domain knowledge using information organizations based on formalisms such as formal ontologies.

Ontologies have gained momentum in robotics. CSIC’s researchers have participated in definition of the 1872–2015 IEEE Standard Ontologies for Robotics and Automation, and the new 1872.2-2021 IEEE Standard for Autonomous Robotics Ontology. We have proposed also OCRA, and ontology to model the knowledge of robots that collaborate with humans and adapt due to unexpected changes in their environment. Furthermore, OCRA has been used to generate robot explanatory narratives of collaborative and adaptive experiences in the line of eXplainable AI (XAI). This enables the robot to detect failures and adaptations during its execution and then narrate what happened to the human to, for example, justify itself or look for solutions.

The objective of the project is to investigate new methods to fill the gap between the reasoning and the execution layers of a robotic system. To fulfill this objective is fundamental to gather in the same team researchers from both subfields in robotics and provide to them space and time to foster cross-fertilization of ideas and generation of new paradigms. The use-case applications proposed are in two very relevant areas: industrial and assistive robotics. The goal in both scenarios is to represent the knowledge so that the robot can better understand how to execute its skills to ensure success, and also to allow robots to reason about when to interact with humans and which content those interactions should contain. It is evident that both potentially have a huge societal and economic impact. At the international level, the project relates to the recent EU AI Act and other initiatives to understand and regulate the use of artificial intelligence and robotics. Because this area is so relevant nowadays, this project will contribute to position both groups internationally and will facilitate the success in international calls devoted not only to scientitific research (low TRLs) but also to technological advances (high TRLs).

Project supported by the call "ACCIONES BILATERALES DE MOVILIDAD CONJUNTA PARA LA PROMOCIÓN DE LA COLABORACIÓN CIENTÍFICA INTERNACIONAL DEL CSIC CON INSTITUCIONES EXTRANJERAS." within the CSIC-NSTC Modality