Search for: All records

Current methods of measuring fairness in human-robot interaction (HRI) research often gauge perceptions of fairness at the conclu- sion of a task. However, this methodology overlooks the dynamic nature of fairness perceptions, which may shift and evolve as a task progresses. To help address this gap, we introduce a platform designed to help investigate the evolution of fairness over time: the Multiplayer Space Invaders game. This three-player game is structured such that two players work to eliminate as many of their own enemies as possible while a third player makes decisions about which player to support throughout the game. In this paper, we discuss different potential experimental designs facilitated by this platform. A key aspect of these designs is the inclusion of a robot that operates the supporting ship and must make multiple decisions about which player to aid throughout a task. We discuss how capturing fairness perceptions at different points in the game could give us deeper insights into how perceptions of fairness fluctuate in response to different variables and decisions made in the game. 

Recent work in Human-Robot Interaction (HRI) has shown that robots can leverage implicit communicative signals from users to understand how they are being perceived during interactions. For example, these signals can be gaze patterns, facial expressions, or body motions that reflect internal human states. To facilitate future research in this direction, we contribute the REACT database, a collection of two datasets of human-robot interactions that display users’ natural reactions to robots during a collaborative game and a photography scenario. Further, we analyze the datasets to show that interaction history is an important factor that can influence human reactions to robots. As a result, we believe that future models for interpreting implicit feedback in HRI should explicitly account for this history. REACT opens up doors to this possibility in the future. 

An overarching goal of Artificial Intelligence (AI) is creating autonomous, social agents that help people. Two important challenges, though, are that different people prefer different assistance from agents and that preferences can change over time. Thus, helping behaviors should be tailored to how an individual feels during the interaction. We hypothesize that human nonverbal behavior can give clues about users' preferences for an agent's helping behaviors, augmenting an agent's ability to computationally predict such preferences with machine learning models. To investigate our hypothesis, we collected data from 194 participants via an online survey in which participants were recorded while playing a multiplayer game. We evaluated whether the inclusion of nonverbal human signals, as well as additional context (e.g., via game or personality information), led to improved prediction of user preferences between agent behaviors compared to explicitly provided survey responses. Our results suggest that nonverbal communication -- a common type of human implicit feedback -- can aid in understanding how people want computational agents to interact with them. 

Recent research in robot learning suggests that implicit human feedback is a low-cost approach to improving robot behavior without the typical teaching burden on users. Because implicit feedback can be difficult to interpret, though, we study different methods to collect fine-grained labels from users about robot performance across multiple dimensions, which can then serve to map implicit human feedback to performance values. In particular, we focused on understanding the effects of annotation order and frequency on human perceptions of the self-annotation process and the usefulness of the labels for creating data-driven models to reason about implicit feedback. Our results demonstrate that different annotation methods can influence perceived memory burden, annotation difficulty, and overall annotation time. Based on our findings, we conclude with recommendations to create future implicit feedback datasets in Human-Robot Interaction. 

Much prior work on creating social agents that assist users relies on preconceived assumptions of what it means to be helpful. For example, it is common to assume that a helpful agent just assists with achieving a user’s objective. However, as assistive agents become more widespread, human-agent interactions may be more ad-hoc, providing opportunities for unexpected agent assistance. How would this affect human notions of an agent’s helpfulness? To investigate this question, we conducted an exploratory study (N=186) where participants interacted with agents displaying unexpected, assistive behaviors in a Space Invaders game and we studied factors that may influence perceived helpfulness in these interactions. Our results challenge the idea that human perceptions of the helpfulness of unexpected agent assistance can be derived from a universal, objective definition of help. Also, humans will reciprocate unexpected assistance, but might not always consider that they are in fact helping an agent. Based on our findings, we recommend considering personalization and adaptation when designing future assistive behaviors for prosocial agents that may try to help users in unexpected situations. 

While neural network binary classifiers are often evaluated on metrics such as Accuracy and F1-Score, they are commonly trained with a cross-entropy objective. How can this training-evaluation gap be addressed? While specific techniques have been adopted to optimize certain confusion matrix based metrics, it is challenging or impossible in some cases to generalize the techniques to other metrics. Adversarial learning approaches have also been proposed to optimize networks via confusion matrix based metrics, but they tend to be much slower than common training methods. In this work, we propose a unifying approach to training neural network binary classifiers that combines a differentiable approximation of the Heaviside function with a probabilistic view of the typical confusion matrix values using soft sets. Our theoretical analysis shows the benefit of using our method to optimize for a given evaluation metric, such as F1-Score, with soft sets, and our extensive experiments show the effectiveness of our approach in several domains.