Towards Trusted Human-Like Artificial Teammates

Today's autonomous systems are still at the level of tools, not that of human-like teammates. In this project, we are developing algorithms and architectures that lay the foundations for future autonomous systems in mixed-initiative human-machine teams to be able to operate at human-like levels of interactivity and effectiveness. We are integrating measurements of neurophysiological signals from human teammates with multiple additional measurements and situational contextual information to classify various individual and team cognitive states. Classified cognitive states of all teammates are individually tracked and fused in real-time, and integrated into a shared mental model which uses advanced probabilistic "theory of mind" representations to capture team and task states with their associated uncertainties, and supports the decision- making and behavior adaptations of the autonomous artificial teammates.

Trade-off between reconstruction loss and feature alignment for domain generalization (2022)

Thuan Nguyen, Boyang Lyu, Prakash Ishwar, Matthias Scheutz, Shuchin Aeron

To deal with challenging settings in domain generalization (DG) where both data and label of the unseen domain are not available at training time, the most common approach is to design the classifiers based on the domain-invariant representation features, i.e., the latent representations that are unchanged and transferable between domains.

Keywords: domain generalization

Joint covariate-alignment and concept-alignment: a framework for domain generalization (2022)

Thuan Nguyen, Boyang Lyu, Prakash Ishwar, Matthias Scheutz, Shuchin Aeron

In this paper, we propose a novel domain generalization (DG) framework based on a new upper bound to the risk on the unseen domain. Particularly, our framework proposes to jointly minimize both the covariate-shift as well as the concept-shift between the seen domains for a better performance on the unseen domain. While the proposed…

Keywords: domain generalization

Conditional entropy minimization principle for learning domain invariant representation features (2022)

Thuan Nguyen, Boyang Lyu, Prakash Ishwar, Matthias Scheutz, Shuchin Aeron

Invariance-principle-based methods such as Invariant Risk Minimization (IRM), have recently emerged as promising approaches for Domain Generalization (DG). Despite promising theory, such approaches fail in common classification tasks due to mixing of "true invariant features" and "spurious invariant features". To address this, we propose…

Keywords: domain generalization

Transparency through Explanations and Justifications in Human-Robot Task-Based Communications (2022)

Scheutz, Matthias and Thielstrom, Ravenna and Abrams, Mitchell

Transparent task-based communication between human instructors and robot instructees requires robots to be able to determine whether a human instruction can and should be carried out, i.e., whether the human is authorized, and whether the robot can and should do it. If the instruction is not appropriate, the robot needs to be able to…

Keywords: human-robot interaction, transparency, explanation, justification

Cognitive Workload Assessment via Eye Gaze and EEG in an Interactive Multi-Modal Driving Task (2022)

Aygun, Ayca and Lyu, Boyang and Nguyen, Thuan and Haga, Zachary and Aeron, Shuchin and Scheutz, Matthias

Assessing cognitive workload of human interactants in mixed initiative teams is a critical capability for autonomous interactive systems to enable adaptations that improve team performance. Yet, it is still unclear, due to diverg ing evidence, which sensing modality might work best for the determination of human workload. In this paper,…

Keywords: workload detection, cognitive load, EEG, eye gaze, driving experiment

Investigating Methods for Cognitive Workload Estimation for Assistive Robots (2022)

Aygun, Ayca and Nguyen, Thuan and Haga, Zachary and Aeron, Shuchin and Scheutz, Matthias

We analyzed and modeled data from a multi-modal simulated driving study specifically designed to evaluate different levels of cognitive workload induced by various secondary tasks such as dialogue interactions and braking events in addition to the primary driving task. Our analyses provide evidence for eye gaze being the best…

Keywords: workload detection, cognitive load, EEG, eye gaze, driving experiment

Mind Readers and Mind Users: The Utility of Sharing Architectural Components across Multiple Robots (2021)

Matthias Scheutz

We introduce the concept of architectural "component-sharing" as the basis for "knowledge sharing" in hive minds, e.g., a system multiple robots connecting by shared components in their control architectures. We discuss the architectural requirements and demonstrate the utility for multi-robot instruction and automatic reasoning across…

Can you trust your trust measure? (2021)

Meia Chita-Tegmark and Theresa Law and Nicholas Rabb and Matthias Scheutz

We examined how typical trust questionnaires used in HRI were affected when participants had the option to choose "not applicable to this robot" or "not applicable to robots in general" for any given question. We found that participants do make use of these choices, particularly for questionnaires that get at social dimensions of trust.

Keywords: Trust, trust measurement, questionnaire design

Enabling Fast Instruction-Based Modification of Learned Robot Skills (2021)

Tyler Frasca and Bradley Oosterveld and Meia Chita-Tegmark and Matthias Scheutz

We develop a framework to allow humans to modify a robot's learned skills through natural language instructions. Additionally, we provide an online study showing that participants would prefer to know they can modify a robot's skills primarily through natural language.

Keywords: Instruction Based Task Learning, Human-Robot Interaction

Trust: Recent Concepts and Evaluations in Human-Robot Interaction (2020)

Theresa Law and Matthias Scheutz

We present a survey of the current empirical literature on trust in HRI. We categorize trust as being either performance-based or relation-based. We compare how each paper in our survey uses performance- or relation-based trust in their trust definitions, research questions, and trust measurements.

Keywords: trust, literature review, performance trust, relational trust

Toward Genuine Robot Teammates: Improving Human-Robot Team Performance Using Robot Shared Mental Models (2020)

Felix Gervits and Dean Thurston and Ravenna Thielstrom and Terry Fong and Quinn Pham and Matthias Scheutz

We implemented a computational framework for Shared Mental Models (SMMs) in which robots use a distributed knowledgebase to coordinate activity. We also built a novel system connecting the robotic architecture, DIARC, to the 3D simulation environment, Unity, to serve as an evaluation platform for the framework implementation.

Keywords: hri, mixed, multi, arch

Simultaneous Representation of Knowledge and Belief for Epistemic Planning with Belief Revision (2020)

David Buckingham and Daniel Kasenberg and Matthias Scheutz

We propose a novel approach to the problem of false belief revision in epistemic planning. Our state representations are pointed Kripke models with two binary relations over possible worlds: one representing agents' necessarily true knowledge, and one representing agents' possibly false beliefs. State transition functions maintain S5n…

Keywords: hri, multi

Generating Explanations for Temporal Logic Planner Decisions (2020)

Daniel Kasenberg and Ravenna Thielstrom and Matthias Scheutz

Although temporal logic has been touted as a fruitful language for specifying interpretable agent objectives, there has been little emphasis on generating explanations for agents with temporal logic objectives. In this paper, we develop an approach to generating explanations for the behavior of agents planning with several temporal logic objectives.

Keywords: moral, comm

Generating Explanations of Action Failures in a Cognitive Robotic Architecture (2020)

Ravenna Thielstrom and Antonio Roque and Meia Chita-Tegmark and Matthias Scheutz

We describe an approach to generating explanations about why robot actions fail, focusing on the considerations of robots that are run by cognitive robotic architectures. We define a set of Failure Types and Explanation Templates, motivating them by the needs and constraints of cognitive architectures that use action scripts and…

Keywords: hri, dia, prag