Search

HRI ’24, March 11–14, 2024, Boulder, CO, USA, In this paper, we introduce a novel conceptual model for a robot's behavioral adaptation in its long-term interaction with humans, integrating dynamic robot role adaptation with principles of flow experience from psychology. This conceptualization introduces a hierarchical interaction objective grounded in the flow experience, serving as the overarching adaptation goal for the robot. This objective intertwines both cognitive and affective sub-objectives and incorporates individual and group-level human factors. The dynamic role adaptation approach is a cornerstone of our model, highlighting the robot's ability to fluidly adapt its support roles-from leader to follower-with the aim of maintaining equilibrium between activity challenge and user skill, thereby fostering the user's optimal flow experiences. Moreover, this work delves into a comprehensive exploration of the limitations and potential applications of our proposed conceptualization. Our model places a particular emphasis on the multi-person HRI paradigm, a dimension of HRI that is both under-explored and challenging. In doing so, we aspire to extend the applicability and relevance of our conceptualization within the HRI field, contributing to the future development of adaptive social robots capable of sustaining long-term interactions with humans.

We describe an approach for aligning an LLM-based dialogue agent based on global (i.e., dialogue-level) rewards, while also taking into account naturally-occurring multimodal signals. At a high level, our approach (dubbed GELI) learns a local, turn-level reward model by decomposing the human-provided Global Explicit (GE) session-level reward, using Local Implicit (LI) multimodal reward signals to crossmodally shape the reward decomposition step. This decomposed reward model is then used as part of the standard RHLF pipeline improve an LLM-based dialog agent. We run quantitative and qualitative human studies to evaluate the performance of our GELI approach, and find that it shows consistent improvements across various conversational metrics compared to baseline methods., Comment: 10 pages, 3 figures, 2 tables

Large language models (LLMs) are capable of many natural language tasks, yet they are far from perfect. In health applications, grounding and interpreting domain-specific and non-linguistic data is crucial. This paper investigates the capacity of LLMs to make inferences about health based on contextual information (e.g. user demographics, health knowledge) and physiological data (e.g. resting heart rate, sleep minutes). We present a comprehensive evaluation of 12 state-of-the-art LLMs with prompting and fine-tuning techniques on four public health datasets (PMData, LifeSnaps, GLOBEM and AW_FB). Our experiments cover 10 consumer health prediction tasks in mental health, activity, metabolic, and sleep assessment. Our fine-tuned model, HealthAlpaca exhibits comparable performance to much larger models (GPT-3.5, GPT-4 and Gemini-Pro), achieving the best performance in 8 out of 10 tasks. Ablation studies highlight the effectiveness of context enhancement strategies. Notably, we observe that our context enhancement can yield up to 23.8% improvement in performance. While constructing contextually rich prompts (combining user context, health knowledge and temporal information) exhibits synergistic improvement, the inclusion of health knowledge context in prompts significantly enhances overall performance.

In this paper, we introduce a novel conceptual model for a robot's behavioral adaptation in its long-term interaction with humans, integrating dynamic robot role adaptation with principles of flow experience from psychology. This conceptualization introduces a hierarchical interaction objective grounded in the flow experience, serving as the overarching adaptation goal for the robot. This objective intertwines both cognitive and affective sub-objectives and incorporates individual and group-level human factors. The dynamic role adaptation approach is a cornerstone of our model, highlighting the robot's ability to fluidly adapt its support roles - from leader to follower - with the aim of maintaining equilibrium between activity challenge and user skill, thereby fostering the user's optimal flow experiences. Moreover, this work delves into a comprehensive exploration of the limitations and potential applications of our proposed conceptualization. Our model places a particular emphasis on the multi-person HRI paradigm, a dimension of HRI that is both under-explored and challenging. In doing so, we aspire to extend the applicability and relevance of our conceptualization within the HRI field, contributing to the future development of adaptive social robots capable of sustaining long-term interactions with humans.

Empathy serves as a cornerstone in enabling prosocial behaviors, and can be evoked through sharing of personal experiences in stories. While empathy is influenced by narrative content, intuitively, people respond to the way a story is told as well, through narrative style. Yet the relationship between empathy and narrative style is not fully understood. In this work, we empirically examine and quantify this relationship between style and empathy using LLMs and large-scale crowdsourcing studies. We introduce a novel, theory-based taxonomy, HEART (Human Empathy and Narrative Taxonomy) that delineates elements of narrative style that can lead to empathy with the narrator of a story. We establish the performance of LLMs in extracting narrative elements from HEART, showing that prompting with our taxonomy leads to reasonable, human-level annotations beyond what prior lexicon-based methods can do. To show empirical use of our taxonomy, we collect a dataset of empathy judgments of stories via a large-scale crowdsourcing study with N=2,624 participants. We show that narrative elements extracted via LLMs, in particular, vividness of emotions and plot volume, can elucidate the pathways by which narrative style cultivates empathy towards personal stories. Our work suggests that such models can be used for narrative analyses that lead to human-centered social and behavioral insights.

Modeling empathy is a complex endeavor that is rooted in interpersonal and experiential dimensions of human interaction, and remains an open problem within AI. Existing empathy datasets fall short in capturing the richness of empathy responses, often being confined to in-lab or acted scenarios, lacking longitudinal data, and missing self-reported labels. We introduce a new multimodal dataset for empathy during personal experience sharing: the EmpathicStories++ dataset (https://mitmedialab.github.io/empathic-stories-multimodal/) containing 53 hours of video, audio, and text data of 41 participants sharing vulnerable experiences and reading empathically resonant stories with an AI agent. EmpathicStories++ is the first longitudinal dataset on empathy, collected over a month-long deployment of social robots in participants' homes, as participants engage in natural, empathic storytelling interactions with AI agents. We then introduce a novel task of predicting individuals' empathy toward others' stories based on their personal experiences, evaluated in two contexts: participants' own personal shared story context and their reflections on stories they read. We benchmark this task using state-of-the-art models to pave the way for future improvements in contextualized and longitudinal empathy modeling. Our work provides a valuable resource for further research in developing empathetic AI systems and understanding the intricacies of human empathy within genuine, real-world settings., Comment: Accepted to ACL 2024 Findings

Foundation models have become invaluable in advancing the medical field. Despite their promise, the strategic deployment of LLMs for effective utility in complex medical tasks remains an open question. Our novel framework, Medical Decision-making Agents (MDAgents) aims to address this gap by automatically assigning the effective collaboration structure for LLMs. Assigned solo or group collaboration structure is tailored to the complexity of the medical task at hand, emulating real-world medical decision making processes. We evaluate our framework and baseline methods with state-of-the-art LLMs across a suite of challenging medical benchmarks: MedQA, MedMCQA, PubMedQA, DDXPlus, PMC-VQA, Path-VQA, and MedVidQA, achieving the best performance in 5 out of 7 benchmarks that require an understanding of multi-modal medical reasoning. Ablation studies reveal that MDAgents excels in adapting the number of collaborating agents to optimize efficiency and accuracy, showcasing its robustness in diverse scenarios. We also explore the dynamics of group consensus, offering insights into how collaborative agents could behave in complex clinical team dynamics. Our code can be found at https://github.com/mitmedialab/MDAgents.

The most meaningful connections between people are often fostered through expression of shared vulnerability and emotional experiences in personal narratives. We introduce a new task of identifying similarity in personal stories based on empathic resonance, i.e., the extent to which two people empathize with each others' experiences, as opposed to raw semantic or lexical similarity, as has predominantly been studied in NLP. Using insights from social psychology, we craft a framework that operationalizes empathic similarity in terms of three key features of stories: main events, emotional trajectories, and overall morals or takeaways. We create EmpathicStories, a dataset of 1,500 personal stories annotated with our empathic similarity features, and 2,000 pairs of stories annotated with empathic similarity scores. Using our dataset, we fine-tune a model to compute empathic similarity of story pairs, and show that this outperforms semantic similarity models on automated correlation and retrieval metrics. Through a user study with 150 participants, we also assess the effect our model has on retrieving stories that users empathize with, compared to naive semantic similarity-based retrieval, and find that participants empathized significantly more with stories retrieved by our model. Our work has strong implications for the use of empathy-aware models to foster human connection and empathy between people., Comment: Published at EMNLP 2023

Accurately modeling affect dynamics, which refers to the changes and fluctuations in emotions and affective displays during human conversations, is crucial for understanding human interactions. By analyzing affect dynamics, we can gain insights into how people communicate, respond to different situations, and form relationships. However, modeling affect dynamics is challenging due to contextual factors, such as the complex and nuanced nature of interpersonal relationships, the situation, and other factors that influence affective displays. To address this challenge, we propose a Cross-person Memory Transformer (CPM-T) framework which is able to explicitly model affective dynamics (intrapersonal and interpersonal influences) by identifying verbal and non-verbal cues, and with a large language model to utilize the pre-trained knowledge and perform verbal reasoning. The CPM-T framework maintains memory modules to store and update the contexts within the conversation window, enabling the model to capture dependencies between earlier and later parts of a conversation. Additionally, our framework employs cross-modal attention to effectively align information from multi-modalities and leverage cross-person attention to align behaviors in multi-party interactions. We evaluate the effectiveness and generalizability of our approach on three publicly available datasets for joint engagement, rapport, and human beliefs prediction tasks. Remarkably, the CPM-T framework outperforms baseline models in average F1-scores by up to 7.3%, 9.3%, and 2.0% respectively. Finally, we demonstrate the importance of each component in the framework via ablation studies with respect to multimodal temporal behavior.

As we move closer to real-world AI systems, AI agents must be able to deal with multiparty (group) conversations. Recognizing and interpreting multiparty behaviors is challenging, as the system must recognize individual behavioral cues, deal with the complexity of multiple streams of data from multiple people, and recognize the subtle contingent social exchanges that take place amongst group members. To tackle this challenge, we propose the Multiparty-Transformer (Multipar-T), a transformer model for multiparty behavior modeling. The core component of our proposed approach is the Crossperson Attention, which is specifically designed to detect contingent behavior between pairs of people. We verify the effectiveness of Multipar-T on a publicly available video-based group engagement detection benchmark, where it outperforms state-of-the-art approaches in average F-1 scores by 5.2% and individual class F-1 scores by up to 10.0%. Through qualitative analysis, we show that our Crossperson Attention module is able to discover contingent behavior., Comment: 7 pages, 4 figures, IJCAI

Introduction: Suicide is a leading cause of death around the world, interpolating a huge suffering to the families and communities of the individuals. Such pain and suffering are preventable with early screening and monitoring. However, current suicide risk identification relies on self-disclosure and/or the clinician's judgment.Research question/statement: Therefore, we investigate acoustic and nonverbal behavioral markers that are associated with different levels of suicide risks through a multimodal approach for suicide risk detection.Given the differences in the behavioral dynamics between subregions of facial expressions and body gestures in terms of timespans, we propose a novel region-based multimodal fusion. Methods: We used a newly collected video interview dataset of young Japanese who are at risk of suicide to extract engineered features and deep representations from the speech, regions of the face (i.e., eyes, nose, mouth), regions of the body (i.e., shoulders, arms, legs), as well as the overall combined regions of face and body. Results: The results confirmed that behavioral dynamics differs between regions, where some regions benefit from a shorter timespans, while other regions benefit from longer ones. Therefore, a region-based multimodal approach is more informative in terms of behavioral markers and accounts for both subtle and strong behaviors. Our region-based multimodal results outperformed the single modality, reaching a sample-level accuracy of 96% compared with the highest single modality that reached sample-level accuracy of 80%. Interpretation of the behavioral markers, showed the higher the suicide risk levels, the lower the expressivity, movement and energy observed from the subject. Moreover, the high-risk suicide group express more disgust and contact avoidance, while the low-risk suicide group express self-soothing and anxiety behaviors. Discussion: Even though multimodal analysis is a powerful tool to enhance the model performance an

This paper presents a method for achieving high-speed running of a quadruped robot by considering the actuator torque-speed operating region in reinforcement learning. The physical properties and constraints of the actuator are included in the training process to reduce state transitions that are infeasible in the real world due to motor torque-speed limitations. The gait reward is designed to distribute motor torque evenly across all legs, contributing to more balanced power usage and mitigating performance bottlenecks due to single-motor saturation. Additionally, we designed a lightweight foot to enhance the robot's agility. We observed that applying the motor operating region as a constraint helps the policy network avoid infeasible areas during sampling. With the trained policy, KAIST Hound, a 45 kg quadruped robot, can run up to 6.5 m/s, which is the fastest speed among electric motor-based quadruped robots.

This paper presents a contact-implicit model predictive control (MPC) framework for the real-time discovery of multi-contact motions, without predefined contact mode sequences or foothold positions. This approach utilizes the contact-implicit differential dynamic programming (DDP) framework, merging the hard contact model with a linear complementarity constraint. We propose the analytical gradient of the contact impulse based on relaxed complementarity constraints to further the exploration of a variety of contact modes. By leveraging a hard contact model-based simulation and computation of search direction through a smooth gradient, our methodology identifies dynamically feasible state trajectories, control inputs, and contact forces while simultaneously unveiling new contact mode sequences. However, the broadened scope of contact modes does not always ensure real-world applicability. Recognizing this, we implemented differentiable cost terms to guide foot trajectories and make gait patterns. Furthermore, to address the challenge of unstable initial roll-outs in an MPC setting, we employ the multiple shooting variant of DDP. The efficacy of the proposed framework is validated through simulations and real-world demonstrations using a 45 kg HOUND quadruped robot, performing various tasks in simulation and showcasing actual experiments involving a forward trot and a front-leg rearing motion., Comment: 22 pages, 19 figures, submitted to International Journal of Robotics Research (IJRR)

Control of legged robots is a challenging problem that has been investigated by different approaches, such as model-based control and learning algorithms. This work proposes a novel Imitating and Finetuning Model Predictive Control (IFM) framework to take the strengths of both approaches. Our framework first develops a conventional model predictive controller (MPC) using Differential Dynamic Programming and Raibert heuristic, which serves as an expert policy. Then we train a clone of the MPC using imitation learning to make the controller learnable. Finally, we leverage deep reinforcement learning with limited exploration for further finetuning the policy on more challenging terrains. By conducting comprehensive simulation and hardware experiments, we demonstrate that the proposed IFM framework can significantly improve the performance of the given MPC controller on rough, slippery, and conveyor terrains that require careful coordination of footsteps. We also showcase that IFM can efficiently produce more symmetric, periodic, and energy-efficient gaits compared to Vanilla RL with a minimal burden of reward shaping.

Affect understanding capability is essential for social robots to autonomously interact with a group of users in an intuitive and reciprocal way. However, the challenge of multi-person affect understanding comes from not only the accurate perception of each user's affective state (e.g., engagement) but also the recognition of the affect interplay between the members (e.g., joint engagement) that presents as complex, but subtle, nonverbal exchanges between them. Here we present a novel hybrid framework for identifying a parent-child dyad's joint engagement by combining a deep learning framework with various video augmentation techniques. Using a dataset of parent-child dyads reading storybooks together with a social robot at home, we first train RGB frame- and skeleton-based joint engagement recognition models with four video augmentation techniques (General Aug, DeepFake, CutOut, and Mixed) applied datasets to improve joint engagement classification performance. Second, we demonstrate experimental results on the use of trained models in the robot-parent-child interaction context. Third, we introduce a behavior-based metric for evaluating the learned representation of the models to investigate the model interpretability when recognizing joint engagement. This work serves as the first step toward fully unlocking the potential of end-to-end video understanding models pre-trained on large public datasets and augmented with data augmentation and visualization techniques for affect recognition in the multi-person human-robot interaction in the wild.

We propose a novel state estimator for legged robots, STEP, achieved through a novel preintegrated foot velocity factor. In the preintegrated foot velocity factor, the usual non-slip assumption is not adopted. Instead, the end effector velocity becomes observable by exploiting the body speed obtained from a stereo camera. In other words, the preintegrated end effector's pose can be estimated. Another advantage of our approach is that it eliminates the necessity for a contact detection step, unlike the typical approaches. The proposed method has also been validated in harsh-environment simulations and real-world experiments containing uneven or slippery terrains.

Across a wide variety of domains, artificial agents that can adapt and personalize to users have potential to improve and transform how social services are provided. Because of the need for personalized interaction data to drive this process, long-term (or longitudinal) interactions between users and agents, which unfold over a series of distinct interaction sessions, have attracted substantial research interest. In recognition of the expanded scope and structure of a long-term interaction, researchers are also adjusting the personalization models and algorithms used, orienting toward “continual learning” methods, which do not assume a stationary modeling target and explicitly account for the temporal context of training data. In parallel, researchers have also studied the effect of “multitask personalization,” an approach in which an agent interacts with users over multiple different tasks contexts throughout the course of a long-term interaction and learns personalized models of a user that are transferrable across these tasks. In this paper, we unite these two paradigms under the framework of “Lifelong Personalization,” analyzing the effect of multitask personalization applied to dynamic, non-stationary targets. We extend the multi-task personalization approach to the more complex and realistic scenario of modeling dynamic learners over time, focusing in particular on interactive scenarios in which the modeling agent plays an active role in teaching the student whose knowledge the agent is simultaneously attempting to model. Inspired by the way in which agents use active learning to select new training data based on domain context, we augment a Gaussian Process-based multitask personalization model with a mechanism to actively and continually manage its own training data, allowing a modeling agent to remove or reduce the weight of observed data from its training set, based on interactive context cues. We evaluate this method in

Despite the increase in awareness and support for mental health, college students’ mental health is reported to decline every year in many countries. Several interactive technologies for mental health have been proposed and are aiming to make therapeutic service more accessible, but most of them only provide one-way passive contents for their users, such as psycho-education, health monitoring, and clinical assessment. We present a robotic coach that not only delivers interactive positive psychology interventions but also provides other useful skills to build rapport with college students. Results from our on-campus housing deployment feasibility study showed that the robotic intervention showed significant association with increases in students’ psychological well-being, mood, and motivation to change. We further found that students’ personality traits were associated with the intervention outcomes as well as their working alliance with the robot and their satisfaction with the interventions. Also, students’ working alliance with the robot was shown to be associated with their pre-to-post change in motivation for better well-being. Analyses on students’ behavioral cues showed that several verbal and nonverbal behaviors were associated with the change in self-reported intervention outcomes. The qualitative analyses on the post-study interview suggest that the robotic coach’s companionship made a positive impression on students, but also revealed areas for improvement in the design of the robotic coach. Results from our feasibility study give insight into how learning users’ traits and recognizing behavioral cues can help an AI agent provide personalized intervention experiences for better mental health outcomes

As voice-user interfaces (VUIs), such as smart speakers like Amazon Alexa or social robots like Jibo, enter multi-user environments like our homes, it is critical to understand how group members perceive and interact with these devices. VUIs engage socially with users, leveraging multi-modal cues including speech, graphics, expressive sounds, and movement. The combination of these cues can affect how users perceive and interact with these devices. Through a set of three elicitation studies, we explore family interactions (N = 34 families, 92 participants, ages 4–69) with three commercially available VUIs with varying levels of social embodiment. The motivation for these three studies began when researchers noticed that families interacted differently with three agents when familiarizing themselves with the agents and, therefore, we sought to further investigate this trend in three subsequent studies designed as a conceptional replication study. Each study included three activities to examine participants’ interactions with and perceptions of the three VUIS in each study, including an agent exploration activity, perceived personality activity, and user experience ranking activity. Consistent for each study, participants interacted significantly more with an agent with a higher degree of social embodiment, i.e., a social robot such as Jibo, and perceived the agent as more trustworthy, having higher emotional engagement, and having higher companionship. There were some nuances in interaction and perception with different brands and types of smart speakers, i.e., Google Home versus Amazon Echo, or Amazon Show versus Amazon Echo Spot between the studies. In the last study, a behavioral analysis was conducted to investigate interactions between family members and with the VUIs, revealing that participants interacted more with the social robot and interacted more with their family members around the interactions with the social robot. This paper

In this work, a non-gaited framework for legged system locomotion is presented. The approach decouples the gait sequence optimization by considering the problem as a decision-making process. The redefined contact sequence problem is solved by utilizing a Monte Carlo Tree Search (MCTS) algorithm that exploits optimization-based simulations to evaluate the best search direction. The proposed scheme has proven to have a good trade-off between exploration and exploitation of the search space compared to the state-of-the-art Mixed-Integer Quadratic Programming (MIQP). The model predictive control (MPC) utilizes the gait generated by the MCTS to optimize the ground reaction forces and future footholds position. The simulation results, performed on a quadruped robot, showed that the proposed framework could generate known periodic gait and adapt the contact sequence to the encountered conditions, including external forces and terrain with unknown and variable properties. When tested on robots with different layouts, the system has also shown its reliability.

© 2020 Owner/Author. A Facebook Messenger chatbot, Sunny, was designed and deployed to promote positive social connections and enhance psychological wellbeing. A 10-day study was conducted with three pre-existing social groups of four members each in control (n=12) and experimental groups (n=12). Both groups completed initial assessments and daily reports, and the experimental groups interacted with Sunny. Exit interviews indicated three key themes: 1) Sunny prompted self-reflection, boosting participants' senses of self-worth and the depth of their relationships, 2) using Sunny encouraged participants to send heartfelt messages they would not have shared otherwise, 3) participants enjoyed accessing positive messages "on-demand". Experimental groups showed an average increase in psychological wellbeing of 1.73 (std = 6.08), compared to 0.5 (std = 5.94) in control. Our results suggest that an AI-based chatbot like Sunny could provide preventative care, promoting strong social connections and psychological wellbeing.

© 2020 IEEE. Conversational AI agents are proliferating, embodying a range of devices such as smart speakers, smart displays, robots, cars, and more. We can envision a future where a personal conversational agent could migrate across different form factors and environments to always accompany and assist its user to support a far more continuous, personalized and collaborative experience. This opens the question of what properties of a conversational AI agent migrates across forms, and how it would impact user perception. To explore this, we developed a Migratable AI system where a user's information and/or the agent's identity can be preserved as it migrates across form factors to help its user with a task. We validated the system by designing a 2x2 between-subjects study to explore the effects of information migration and identity migration on user perceptions of trust, competence, likeability and social presence. Our results suggest that identity migration had a positive effect on trust, competence and social presence, while information migration had a positive effect on trust, competence and likeability. Overall, users report highest trust, competence, likeability and social presence towards the conversational agent when both identity and information were migrated across embodiments.

© 2019 International Joint Conferences on Artificial Intelligence. All rights reserved. Intelligent tutoring systems (ITS) provide educational benefits through one-on-one tutoring by assessing children's existing knowledge and providing tailored educational content. In the domain of language acquisition, several studies have shown that children often learn new words by forming semantic relationships with words they already know. In this paper, we present a model that uses word semantics (semantics-based model) to make inferences about a child's vocabulary from partial information about their existing vocabulary knowledge. We show that the proposed semantics-based model outperforms models that do not use word semantics (semantics-free models) on average. A subject-level analysis of results reveals that different models perform well for different children, thus motivating the need to combine predictions. To this end, we use two methods to combine predictions from semantics-based and semantics-free models and show that these methods yield better predictions of a child's vocabulary knowledge. Our results motivate the use of semantics-based models to assess children's vocabulary knowledge and build ITS that maximizes children's semantic understanding of words., National Science Foundation (Grant IIS-1734443)

© 2019, Association for the Advancement of Artificial Intelligence (www.aaai.org). All rights reserved. PopBots is a hands-on toolkit and curriculum designed to help young children learn about artificial intelligence (AI) by building, programming, training, and interacting with a social robot. Today's children encounter AI in the forms of smart toys and computationally curated educational and entertainment content. However, children have not yet been empowered to understand or create with this technology. Existing computational thinking platforms have made ideas like sequencing and conditionals accessible to young learners. Going beyond this, we seek to make AI concepts accessible. We designed PopBots to address the specific learning needs of children ages four to seven by adapting constructionist ideas into an AI curriculum. This paper describes how we designed the curriculum and evaluated its effectiveness with 80 Pre-K and Kindergarten children. We found that the use of a social robot as a learning companion and programmable artifact was effective in helping young children grasp AI concepts. We also identified teaching approaches that had the greatest impact on student's learning. Based on these, we make recommendations for future modules and iterations for the PopBots platform.

© 2019 IEEE. Perceiving users' engagement accurately is important for technologies that need to respond to learners in a natural and intelligent way. In this paper, we address the problem of automated estimation of engagement from videos of child-robot interactions recorded in unconstrained environments (kindergartens). This is challenging due to diverse and person-specific styles of engagement expressions through facial and body gestures, as well as because of illumination changes, partial occlusion, and a changing background in the classroom as each child is active. To tackle these difficult challenges, we propose a novel deep reinforcement learning architecture for active learning and estimation of engagement from video data. The key to our approach is the learning of a personalized policy that enables the model to decide whether to estimate the child's engagement level (low, medium, high) or, when uncertain, to query a human for a video label. Queried videos are labeled by a human expert in an offline manner, and used to personalize the policy and engagement classifier to a target child over time. We show on a database of 43 children involved in robot-assisted learning activities (8 sessions over 3 months), that this combined human-AI approach can easily adapt its interpretations of engagement to the target child using only a handful of labeled videos, while being robust to the many complex influences on the data. The results show large improvements over a non-personalized approach and over traditional active learning methods.

Can robots help children be more creative? In this work, we posit social robots as creativity support tools for children in collaborative interactions. Children learn creative expressions and behaviors through social interactions with others during playful and collaborative tasks, and socially emulate their peers’ and teachers’ creativity. Social robots have a unique ability to engage in social and emotional interactions with children that can be leveraged to foster creative expression. We focus on two types of social interactions: creativity demonstration, where the robot exhibits creative behaviors, and creativity scaffolding, where the robot poses challenges, suggests ideas, provides positive reinforcement, and asks questions to scaffold children’s creativity. We situate our research in three playful and collaborative tasks - the Droodle Creativity game (that affords verbal creativity), the MagicDraw game (that affords figural creativity), and the WeDo construction task (that affords constructional creativity), that children play with Jibo, a social robot. To evaluate the efficacy of the robot’s social behaviors in enhancing creative behavior and expression in children, we ran three randomized controlled trials with 169 children in the 5–10 yr old age group. In the first two tasks, the robot exhibited creativity demonstration behaviors. We found that children who interacted with the robot exhibiting high verbal creativity in the Droodle game and high figural creativity in the MagicDraw game also exhibited significantly higher creativity than a control group of participants who interacted with a robot that did not express creativity (p < 0.05*). In the WeDo construction task, children who interacted with the robot that expressed creative scaffolding behaviors (asking reflective questions, generating ideas and challenges, and providing positive reinforcement) demonstrated higher creativity than participants in the control group by expressing a greater number of ide

© 2019 Association for Computing Machinery. We developed a novel early childhood artificial intelligence (AI) platform, PopBots, where preschool children train and interact with social robots to learn three AI concepts: knowledge-based systems, supervised machine learning, and generative AI. We evaluated how much children learned by using AI assessments we developed for each activity. The median score on the cumulative assessment was 70% and children understood knowledge-based systems the best. Then, we analyzed the impact of the activities on children’s perceptions of robots. Younger children came to see robots as toys that were smarter than them, but their older counterparts saw them more as people that were not as smart as them. Children who performed worse on the AI assessments believed that robots were like toys that were not as smart as them, however children who did better on the assessments saw robots as people who were smarter than them. We believe early AI education can empower children to understand the AI devices that are increasingly in their lives.

The storytelling lens in human-computer interaction has primarily focused on personas, design fiction, and other stories crafted by designers, yet informal personal narratives from everyday people have not been considered meaningful data, such as storytelling from older adults. Storytelling may provide a clear path to conceptualize how technologies such as social robots can support the lives of older or disabled individuals. To explore this, we engaged 28 older adults in a year-long co-design process, examining informal stories told by older adults as a means of generating and expressing technology ideas and needs. This paper presents an analysis of participants’ stories around their prior experience with technology, stories shaped by social context, and speculative scenarios for the future of social robots. From this analysis, we present suggestions for social robot design, considerations of older adults’ values around technology design, and promotion of participant stories as sources for design knowledge and shifting perspectives of older adults and technology.

© 2019, Association for the Advancement of Artificial Intelligence (www.aaai.org). All rights reserved. Personalized education technologies capable of delivering adaptive interventions could play an important role in addressing the needs of diverse young learners at a critical time of school readiness. We present an innovative personalized social robot learning companion system that utilizes children's verbal and nonverbal affective cues to modulate their engagement and maximize their long-term learning gains. We propose an affective reinforcement learning approach to train a personalized policy for each student during an educational activity where a child and a robot tell stories to each other. Using the personalized policy, the robot selects stories that are optimized for each child's engagement and linguistic skill progression. We recruited 67 bilingual and English language learners between the ages of 4-6 years old to participate in a between-subjects study to evaluate our system. Over a three-month deployment in schools, a unique storytelling policy was trained to deliver a personalized story curriculum for each child in the Personalized group. We compared their engagement and learning outcomes to a Non-personalized group with a fixed curriculum robot, and a baseline group that had no robot intervention. In the Personalization condition, our results show that the affective policy successfully personalized to each child to boost their engagement and outcomes with respect to learning and retaining more target words as well as using more target syntax structures as compared to children in the other groups.

Across a wide variety of domains, artificial agents that can adapt and personalize to users have potential to improve and transform how social services are provided. Because of the need for personalized interaction data to drive this process, long-term (or longitudinal) interactions between users and agents, which unfold over a series of distinct interaction sessions, have attracted substantial research interest. In recognition of the expanded scope and structure of a long-term interaction, researchers are also adjusting the personalization models and algorithms used, orienting toward “continual learning” methods, which do not assume a stationary modeling target and explicitly account for the temporal context of training data. In parallel, researchers have also studied the effect of “multitask personalization,” an approach in which an agent interacts with users over multiple different tasks contexts throughout the course of a long-term interaction and learns personalized models of a user that are transferrable across these tasks. In this paper, we unite these two paradigms under the framework of “Lifelong Personalization,” analyzing the effect of multitask personalization applied to dynamic, non-stationary targets. We extend the multi-task personalization approach to the more complex and realistic scenario of modeling dynamic learners over time, focusing in particular on interactive scenarios in which the modeling agent plays an active role in teaching the student whose knowledge the agent is simultaneously attempting to model. Inspired by the way in which agents use active learning to select new training data based on domain context, we augment a Gaussian Process-based multitask personalization model with a mechanism to actively and continually manage its own training data, allowing a modeling agent to remove or reduce the weight of observed data from its training set, based on interactive context cues. We evaluate this method in

© 2018 Association for Computing Machinery. Social robots are increasingly being developed for long-term interactions with children in domains such as healthcare, education, therapy, and entertainment. As such, we need to deeply understand how children's relationships with robots develop through time. However, there are few validated assessments for measuring young children's long-term relationships. In this paper, we present a pilot test of four assessments that we have adapted or created for use in this context with children aged 5-6: the Inclusion of Other in Self task, the Social-Relational Interview, the Narrative Description, and the Self-disclosure Task. We show that children can appropriately respond to these assessments with reasonably high internal reliability, and that the proposed assessments are able to capture child-robot relationship adjustments over a long-term interaction. Furthermore, we discuss gender and population differences in children's responses.

© 2020 Elsevier Ltd Pedagogical agents are typically designed to take on a single role: either as a tutor who guides and instructs the student, or as a tutee that learns from the student to reinforce what he/she knows. While both agent-role paradigms have been shown to promote student learning, we hypothesize that there will be heightened benefit with respect to students’ learning and emotional engagement if the agent engages children in a more peer-like way — adaptively switching between tutor/tutee roles. In this work, we present a novel active role-switching (ARS) policy trained using reinforcement learning, in which the agent is rewarded for adapting its tutor or tutee behavior to the child's knowledge mastery level. To investigate how the three different child–agent interaction paradigms (tutee, tutor, and peer agents) impact children's learning and affective engagement, we designed a randomized controlled between-subject experiment. Fifty-nine children aged 5–7 years old from a local public school participated in a collaborative word-learning activity with one of the three agent-role paradigms. Our analysis revealed that children's vocabulary acquisition benefited from the robot tutor's instruction and knowledge demonstration, whereas children exhibited slightly greater affect on their faces when the robot behaves as a tutee of the child. This synergistic effect between tutor and tutee roles suggests why our adaptive peer-like agent brought the most benefit to children's vocabulary learning and affective engagement, as compared to an agent that interacts only as a tutor or tutee for the child. This work sheds light on how fixed role (tutor/tutee) and adaptive role (peer) agents support children's cognitive and emotional needs as they play and learn. It also contributes to an important new dimension of designing educational agents — actively adapting roles based on the student's engagement and learning needs.

© 2020 Association for Computing Machinery. Prior work in affect-aware educational robots has often relied on a common belief that the relationship between student affect and learning is independent of agent behaviors (child's/robot's) or unidirectional (positive/negative but not both) throughout the entire student-robot interaction.We argue that the student affect-learning relationship should be interpreted in two contexts: (1) social learning paradigm and (2) sub-events within child-robot interaction. In our paper, we examine two different social learning paradigms where children interact with a robot that acts either as a tutor or a tutee. Sub-events within child-robot interaction are defined as task-related events occurring in specific phases of an interaction (e.g., when the child/robot gets a wrong answer). We examine subevents at a macro level (entire interaction) and a micro level (within specific sub-events). In this paper, we provide an in-depth correlation analysis of children's facial affect and vocabulary learning. We found that children's affective displays became more predictive of their vocabulary learning when children interacted with a tutee robot who did not scaffold their learning. Additionally, children's affect displayed during micro-level events was more predictive of their learning than during macro-level events. Last, we found that the affect-learning relationship is not unidirectional, but rather is modulated by context, i.e., several affective states facilitated student learning when displayed in some sub-events but inhibited learning when displayed in others. These findings indicate that both social learning paradigm and sub-events within interaction modulate student affect-learning relationship.

© 2019 International Joint Conferences on Artificial Intelligence. All rights reserved. Intelligent tutoring systems (ITS) provide educational benefits through one-on-one tutoring by assessing children's existing knowledge and providing tailored educational content. In the domain of language acquisition, several studies have shown that children often learn new words by forming semantic relationships with words they already know. In this paper, we present a model that uses word semantics (semantics-based model) to make inferences about a child's vocabulary from partial information about their existing vocabulary knowledge. We show that the proposed semantics-based model outperforms models that do not use word semantics (semantics-free models) on average. A subject-level analysis of results reveals that different models perform well for different children, thus motivating the need to combine predictions. To this end, we use two methods to combine predictions from semantics-based and semantics-free models and show that these methods yield better predictions of a child's vocabulary knowledge. Our results motivate the use of semantics-based models to assess children's vocabulary knowledge and build ITS that maximizes children's semantic understanding of words.

Human-robot interaction can be regarded as a flow between users and robots. Designing good interaction flows takes a lot of effort and needs to be field tested. Unfortunately, the interaction flow design process is often very disjointed, with users experiencing prototypes, designers forming those prototypes, and developers implementing them as independent processes. In this paper, we present the Interaction Flow Editor (IFE), a new human-robot interaction prototyping tool that enables everyday users to create and modify their own interactions, while still providing a full suite of features that is powerful enough for developers and designers to create complex interactions. We also discuss the Flow Engine, a flexible and adaptable framework for executing robot interaction flows authors through the IFE. Finally, we present our case study results that demonstrates how older adults, aged 70 and above, can design and iterate interactions in real-time on a robot using the IFE., Comment: 8 pages, 4 figures

Conversational AI agents are proliferating, embodying a range of devices such as smart speakers, smart displays, robots, cars, and more. We can envision a future where a personal conversational agent could migrate across different form factors and environments to always accompany and assist its user to support a far more continuous, personalized, and collaborative experience. This opens the question of what properties of a conversational AI agent migrates across forms, and how it would impact user perception. To explore this, we developed a Migratable AI system where a user's information and/or the agent's identity can be preserved as it migrates across form factors to help its user with a task. We designed a 2x2 between-subjects study to explore the effects of information migration and identity migration on user perceptions of trust, competence, likeability, and social presence. Our results suggest that identity migration had a positive effect on trust, competence, and social presence, while information migration had a positive effect on trust, competence, and likeability. Overall, users report the highest trust, competence, likeability, and social presence towards the conversational agent when both identity and information were migrated across embodiments., Comment: Accepted to RO-MAN 2020

This paper presents a novel Representation-Free Model Predictive Control (RF-MPC) framework for controlling various dynamic motions of a quadrupedal robot in three dimensional (3D) space. Our formulation directly represents the rotational dynamics using the rotation matrix, which liberates us from the issues associated with the use of Euler angles and quaternion as the orientation representations. With a variation-based linearization scheme and a carefully constructed cost function, the MPC control law is transcribed to the standard Quadratic Program (QP) form. The MPC controller can operate at real-time rates of 250 Hz on a quadruped robot. Experimental results including periodic quadrupedal gaits and a controlled backflip validate that our control strategy could stabilize dynamic motions that involve singularity in 3D maneuvers.

This paper proposes a kinodynamic motion planning framework for multi-legged robot jumping based on the mixed-integer convex program (MICP), which simultaneously reasons about centroidal motion, contact points, wrench, and gait sequences. This method uniquely combines configuration space discretization and the construction of feasible wrench polytope (FWP) to encode kinematic constraints, actuator limit, friction cone constraint, and gait sequencing into a single MICP. The MICP could be efficiently solved to the global optimum by off-the-shelf numerical solvers and provide highly dynamic jumping motions without requiring initial guesses. Simulation and experimental results demonstrate that the proposed method could find novel and dexterous maneuvers that are directly deployable on the two-legged robot platform to traverse through challenging terrains.

A significant number of college students suffer from mental health issues that impact their physical, social, and occupational outcomes. Various scalable technologies have been proposed in order to mitigate the negative impact of mental health disorders. However, the evaluation for these technologies, if done at all, often reports mixed results on improving users' mental health. We need to better understand the factors that align a user's attributes and needs with technology-based interventions for positive outcomes. In psychotherapy theory, therapeutic alliance and rapport between a therapist and a client is regarded as the basis for therapeutic success. In prior works, social robots have shown the potential to build rapport and a working alliance with users in various settings. In this work, we explore the use of a social robot coach to deliver positive psychology interventions to college students living in on-campus dormitories. We recruited 35 college students to participate in our study and deployed a social robot coach in their room. The robot delivered daily positive psychology sessions among other useful skills like delivering the weather forecast, scheduling reminders, etc. We found a statistically significant improvement in participants' psychological wellbeing, mood, and readiness to change behavior for improved wellbeing after they completed the study. Furthermore, students' personality traits were found to have a significant association with intervention efficacy. Analysis of the post-study interview revealed students' appreciation of the robot's companionship and their concerns for privacy., Comment: 8 pages, 5 figures, RO-MAN 2020, Best paper award

Automatic speech-based affect recognition of individuals in dyadic conversation is a challenging task, in part because of its heavy reliance on manual pre-processing. Traditional approaches frequently require hand-crafted speech features and segmentation of speaker turns. In this work, we design end-to-end deep learning methods to recognize each person's affective expression in an audio stream with two speakers, automatically discovering features and time regions relevant to the target speaker's affect. We integrate a local attention mechanism into the end-to-end architecture and compare the performance of three attention implementations -- one mean pooling and two weighted pooling methods. Our results show that the proposed weighted-pooling attention solutions are able to learn to focus on the regions containing target speaker's affective information and successfully extract the individual's valence and arousal intensity. Here we introduce and use a "dyadic affect in multimodal interaction - parent to child" (DAMI-P2C) dataset collected in a study of 34 families, where a parent and a child (3-7 years old) engage in reading storybooks together. In contrast to existing public datasets for affect recognition, each instance for both speakers in the DAMI-P2C dataset is annotated for the perceived affect by three labelers. To encourage more research on the challenging task of multi-speaker affect sensing, we make the annotated DAMI-P2C dataset publicly available, including acoustic features of the dyads' raw audios, affect annotations, and a diverse set of developmental, social, and demographic profiles of each dyad., Comment: Accepted by the 2020 International Conference on Multimodal Interaction (ICMI'20)

Mindset has been shown to have a large impact on people's academic, social, and work achievements. A growth mindset, i.e., the belief that success comes from effort and perseverance, is a better indicator of higher achievements as compared to a fixed mindset, i.e., the belief that things are set and cannot be changed. Interventions aimed at promoting a growth mindset in children range from teaching about the brain's ability to learn and change, to playing computer games that grant brain points for effort rather than success. This work explores a novel paradigm to foster a growth mindset in young children where they play a puzzle solving game with a peer-like social robot. The social robot is fully autonomous and programmed with behaviors suggestive of it having either a growth mindset or a neutral mindset as it plays puzzle games with the child. We measure the mindset of children before and after interacting with the peer-like robot, in addition to measuring their problem solving behavior when faced with a challenging puzzle. We found that children who played with a growth-mindset robot 1) self-reported having a stronger growth mindset and 2) tried harder during a challenging task, as compared to children who played with the neutral-mindset robot. These results suggest that interacting with peer-like social robot with a growth mindset can promote the same mindset in children. Keywords: early childhood education; mindset; perseverance; grit; child-robot interaction; cognitive architecture; social robots, National Institute of Health (Grant 5R01HD086899–02)

While there has been a growing body of work in child-robot interaction, we still have very little knowledge regarding young children's speaking and listening dynamics and how a robot companion should decode these behaviors and encode its own in a way children can understand. In developing a backchannel prediction model based on observed nonverbal behaviors of 4-6 year-old children, we investigate the effects of an attentive listening robot on a child's storytelling. We provide an extensive analysis of young children's nonverbal behavior with respect to how they encode and decode listener responses and speaker cues. Through a collected video corpus of peer-to-peer storytelling interactions, we identify attention-related listener behaviors as well as speaker cues that prompt opportunities for listener backchannels. Based on our findings, we developed a backchannel opportunity prediction (BOP) model that detects four main speaker cue events based on prosodic features in a child's speech. This rule-based model is capable of accurately predicting backchanneling opportunities in our corpora. We further evaluate this model in a human-subjects experiment where children told stories to an audience of two robots, each with a different backchanneling strategy. We find that our BOP model produces contingent backchannel responses that conveys an increased perception of an attentive listener, and children prefer telling stories to the BOP model robot., National Science Foundation (U.S.) (NSF grant IIS-1523118)