Trustworthiness Assessment of an Adaptive and Explainable Robot in a Real Environment

Abstract Trust plays a fundamental role in shaping the dynamics of Human-Robot Interaction and it is closely related to the complex mental attitudes of task delegation and task adoption. This work contributes to the existing literature by analyzing results from an experiment conducted in real museum settings. The experiment investigates the relation between robot’s competence, willingness, and trustworthiness, as well as the impact of robot’s actions explanation on the aforementioned dimensions. Specifically, it focuses on an adaptive robot capable of adopting a task delegated by the human at various autonomy levels and explaining the reasons behind its actions. The explanation is based on the human’s mental states (Theory of Mind). Findings demonstrated that an adaptive robot that deviates from the human’s explicitly stated expectations, is perceived as trustworthy if evaluated competent in the collaboration, regardless of the explanation provided. Additionally, a robot evaluated willing to interact with the human for achieving the delegated task is perceived as trustworthy when it is able to explain the mechanism beyond the task adoption process. Furthermore, we found that a human who accepts and comprehends the logic underlying the robot’s explanation tends to evaluate the provided explanation positively. Taken together, these results contribute to assess the relationship between trustworthiness, competence, willingness, and explainability in adaptive robots.