arXiv subjects

Manas Satish Bedmutha

Publications and source records attributed to Manas Satish Bedmutha.

6 recordsLinked to original sources

arXiv2026

AffAdapt: AFFect-driven ADAPTive AI Personas for Seamless Conversations

AI-generated personas are being increasingly used for support, training and simulations. While generative AI models possess abilities to generate affect-aware responses, their embodiment into visual personas is an active area of investigation. Naturalistic exchanges require understanding of the conversational partners' turn completions, whether the agent should respond or keep listening and rely on non-verbal cues aligned with one's emotional states. Seamless human-AI conversation in a multimodal setting requires all modalities being generated to act in coordination. We present AffAdapt, a seamless interaction design framework for AI-personas, which coordinates streaming speech recognition, proactive turn-management, persona-grounded response generation, a persistent emotional state, and synchronized embodied output into a single interaction loop. We demonstrate the architecture in the context of practicing sensitive, high-stakes conversations, and report an initial case study showing fluid turn management and adaptive, persona-consistent behavior, alongside open challenges in interruption handling, open-ended dialogue, and multimodal affective alignment. AffAdapt's interaction loop is a generalizable pattern for coordinating timing, identity, and affect in real-time AI personas - applicable to training, coaching, education, and simulation contexts wherever believable, responsive interaction matters.

cs.HC↗

arXiv2026

SocialLM: Social Signal Processing of Patient-Provider Communication using LLMs and Contextual Aggregation

Effective patient-provider communication is difficult to assess at scale. We examine whether large language models (LLMs) can track 20 social behaviors from clinical transcripts without fine-tuning. Across three model families and multiple prompting strategies, LLMs reliably detect social signals, though performance varies by patient race and visit segment. To address this variability under query-only API constraints, we introduce an agreement-weighted ensemble using group-level agreement patterns. This approach improves both accuracy and stability over the best individual model, demonstrating a practical pathway for scalable social signal tracking in clinical conversations.

cs.CL↗

arXiv2025

DriveSimQuest: A VR Driving Simulator and Research Platform on Meta Quest with Unity

Using head-mounted Virtual Reality (VR) displays to simulate driving is critical to studying driving behavior and designing driver assistance systems. But existing VR driving simulators are often limited to tracking only eye movements. The bulky outside-in tracking setup and Unreal-based architecture also present significant engineering challenges for interaction researchers and practitioners. We present DriveSimQuest, a VR driving simulator and research platform built on the Meta Quest Pro and Unity, capable of capturing rich behavioral signals such as gaze, facial expressions, hand activities, and full-body gestures in real-time. DriveSimQuest offers a preliminary, easy-to-deploy platform that supports researchers and practitioners in studying drivers' affective states and behaviors, and in designing future context-aware driving assistance systems.

cs.HC↗

arXiv2025

Predicting Trust In Autonomous Vehicles: Modeling Young Adult Psychosocial Traits, Risk-Benefit Attitudes, And Driving Factors With Machine Learning

Low trust remains a significant barrier to Autonomous Vehicle (AV) adoption. To design trustworthy AVs, we need to better understand the individual traits, attitudes, and experiences that impact people's trust judgements. We use machine learning to understand the most important factors that contribute to young adult trust based on a comprehensive set of personal factors gathered via survey (n = 1457). Factors ranged from psychosocial and cognitive attributes to driving style, experiences, and perceived AV risks and benefits. Using the explainable AI technique SHAP, we found that perceptions of AV risks and benefits, attitudes toward feasibility and usability, institutional trust, prior experience, and a person's mental model are the most important predictors. Surprisingly, psychosocial and many technology- and driving-specific factors were not strong predictors. Results highlight the importance of individual differences for designing trustworthy AVs for diverse groups and lead to key implications for future design and research.

cs.HC↗

arXiv2024

Toward Automated Detection of Biased Social Signals from the Content of Clinical Conversations

Implicit bias can impede patient-provider interactions and lead to inequities in care. Raising awareness is key to reducing such bias, but its manifestations in the social dynamics of patient-provider communication are difficult to detect. In this study, we used automated speech recognition (ASR) and natural language processing (NLP) to identify social signals in patient-provider interactions. We built an automated pipeline to predict social signals from audio recordings of 782 primary care visits that achieved 90.1% average accuracy across codes, and exhibited fairness in its predictions for white and non-white patients. Applying this pipeline, we identified statistically significant differences in provider communication behavior toward white versus non-white patients. In particular, providers expressed more patient-centered behaviors towards white patients including more warmth, engagement, and attentiveness. Our study underscores the potential of automated tools in identifying subtle communication signals that may be linked with bias and impact healthcare quality and equity.

cs.CY↗

arXiv2023

Towards Enhanced Human Activity Recognition through Natural Language Generation and Pose Estimation

Vision-based human activity recognition (HAR) has made substantial progress in recognizing predefined gestures but lacks adaptability for emerging activities. This paper introduces a paradigm shift by harnessing generative modeling and large language models (LLMs) to enhance vision-based HAR. We propose utilizing LLMs to generate descriptive textual representations of activities using pose keypoints as an intermediate representation. Incorporating pose keypoints adds contextual depth to the recognition process, allowing for sequences of vectors resembling text chunks, compatible with LLMs. This innovative fusion of computer vision and natural language processing holds significant potential for revolutionizing activity recognition. A proof of concept study on a Kinetics700 dataset subset validates the approach's efficacy, highlighting improved accuracy and interpretability. Future implications encompass enhanced accuracy, novel research avenues, model generalization, and ethical considerations for transparency. This framework has real-world applications, including personalized gym workout feedback and nuanced sports training insights. By connecting visual cues to interpretable textual descriptions, the proposed framework advances HAR accuracy and applicability, shaping the landscape of pervasive computing and activity recognition research. As this approach evolves, it promises a more insightful understanding of human activities across diverse contexts, marking a significant step towards a better world.

cs.HC↗