Architecting Explainable Service Robots

Service robots entailing a tight collaboration with humans are increasingly widespread in critical domains, such as healthcare and domestic assistance. However, the so-called Human-Machine-Teaming paradigm can be hindered by the black-box nature of service robots, whose autonomous decisions may be confusing or even dangerous for humans. Thus, the explainability for these systems emerges as a crucial property for their acceptance in our society. This paper introduces the concept of explainable service robots and proposes a software architecture to support the engineering of the self-explainability requirements in these collaborating systems by combining formal analysis and interpretable machine learning. We evaluate the proposed architecture using an illustrative example in healthcare. Results show that our proposal supports the explainability of multi-agent Human-Machine-Teaming missions featuring an infinite (dense) space of human-machine uncertain factors, such as diverse physical and physiological characteristics of the agents involved in the teamwork.