Shared Mental Models to Support Distributed Human-Robot Teaming in Space

Abstract

Effective human-robot interaction (HRI) is a critical requirement for current and future space operations. However, given the limitations of autonomous technologies, robots are not yet capable of coordinating with human crew as peers under real-world mission constraints. Due to the complexity inherent in space robotics operations, it is crucial that robots are able to coordinate the various aspects of human-robot teaming and the task at hand. In this paper, we extend our Shared Mental Model (SMM) interaction framework to show how it can be used to overcome some of the challenges inherent in distributed HRI and facilitate coordination in space robotics teams. Since this framework has not yet been implemented in real systems, we conducted an exploratory study to identify potential benefits that the SMM mechanisms can afford in a task domain involving simulated free-flying robot assistants on a spacecraft. We found that the SMM framework offers advantages to task performance and team efficiency due to its support of shared knowledge representations, however these benefits do not seem to reduce workload or improve other subjective measures. The significance of these findings for future space operations is discussed, as are directions for future research. * Note that true independence may not even be considered interaction by some definitions. We are using the term to refer to tasks in which the interdependent human and robot actions are separated in time.

A Review of NASA Human-Robot Interaction in Space

“…Gervits et al have studied how a shared mental model (SMM) can be applied to human-robot teaming in space, including both co-located and widely distributed teams [39,40]. A SMM is a construct that contains the knowledge and functions necessary for successful teamwork, including monitoring goal and task states, evaluating performance of teammates, inferring beliefs and intentions, tracking task focus, and adapting behavior as a result of this knowledge.…”

Understanding Human Decision Processes: Inferring Decision Strategies From Behavioral Data

“…However, most of these systems have little to no understanding of their human teammate's decision-making process and/or cognitive state, leaving such systems unable to adapt the aid they provide to match the needs of their human teammate. The inability of the AI system to perceive and incorporate information about the human's perspective into their own programming is a crucial missing component of effective information sharing in human-AI teams (Gervits et al, 2018;Scheutz et al, 2017). Artificial intelligence systems must develop approximate models of the human decision process which can be quickly trained and applied toward anticipating human behavior (Chakraborti et al, 2017).…”

Artificial Intelligence and Groups: Effects of Attitudes and Discretion on Collaboration

“…Prior work focused on less intelligent technologies (e.g., fixed automation, algorithms, and tools built with AI but not dynamic in their implementation, c.f., L. Larson & DeChurch, 2020). As noted above, AI in human–AI teams are more than mere tools for information retrieval (Gervits et al, 2020). Instead, we see scenarios where AI can choose or recommend courses of action (Murray et al, 2021; O’Neill et al, 2022).…”

“…In the context of teams, we focus on AI that exhibits some form of self-governance for decision-making, adaptation, and communication (Demir et al, 2016; Mercado et al, 2016; Myers et al, 2019). The role of AI in human–AI teams goes beyond serving as mere tools for information retrieval (Gervits et al, 2020). AI roles extend to scenarios where AI can choose or recommend a course of action that supports the team (Murray et al, 2021; O’Neill et al, 2022).…”