Robots, Listening & Learning Interactions
Date: Tuesday 11 Nov 2025
Room: S/3.20 Turing Suite
| Time | Presentation | Abstract | Authors | Contribution type |
|---|---|---|---|---|
| 14:00-14:10 | Active Listening in Interactive Coaching: Prompt Strategies and User Assessment | For Conversational Agents (CA) to live up to their potential in supporting behaviour change for good, it is essential that high levels of appropriate conversational skills be applied in a systematic and controllable way. In light of this challenge, we present a model and study of the role of Active Listening in virtual health coaches. We detail the design and parameterization of AMOS, an LLM driven virtual health coach, and present a 45 participant within-subject evaluation of the impact of explicit Active Listening behaviour on participant attitudes toward AMOS itself and our targeted behaviour change goals. The results demonstrated that explicit Active Listening resulted in higher engagement and effectiveness in the potential for achieving health goals in comparison to an alternative focused on direct support compared to a chatbot lacking these attributes. Additionally, data on participants’ personality scores were collected to explore potential relationships between personality traits and intervention outcomes. Participant openness and conscientiousness appear to positively influence outcomes in the case of explicit Active Listening, but this effect tends to weaken—or even reverse—in the case of no explicit Active Listening. | Ghulam Hussain, Brian Keegan and Robert Ross | Late Breaking Work |
| 14:10-14:20 | The Benefit of Multi-robot Socially Interactive Systems: A Survey of Controlled Experiments | A survey of 28 human-robot interaction controlled experiments assessed the benefits of multi-agent interactive systems (e.g., two robots) vs. lone robots. Design factors explored included: input, output, agent-agent communication, role, personality, physical appearance, hardware/software embodiment and agent name. The benefits of two or more robots vs. a lone robot based on 16 of 28 comparative studies finding positive effects include greater synchronization, attention, liking of praise, sales and cuteness. Preliminary analysis of design factors grouped by experimental result showed that the presence of agent-agent communication and the same hardware/software may be useful to achieving a benefit of multiple robots. This work offers rationale for using multiple robots in social interaction as well as an exploration of the design space for such multi-robot interaction. | Jamy Li, Aleksander Bielinski and Callum Goddard | Late Breaking Work |
| 14:20-14:30 | Envisioning Social Robots in Daily Life: A Home-Based Design Study with Older Adults in Pakistan | As life expectancy increases, the global population is aging, which has led to increased interest in social robots designed to support independent living, emotional well-being, and caregiving. However, most research and development in this area is based on Western cultural assumptions and infrastructures, creating a gap in understanding how these social robots may be accepted in low- and middle-income countries (LMICs). In this paper, we present findings from our design sessions conducted with older adults in Pakistan, as part of a broader home-based research study exploring the role of social robots in later stages of life. This study examined how older adults envision incorporating robots into their daily routines and what expectations they have regarding their use through co-design methods. The results highlight the importance of culturally sensitive robot design and user involvement in the development of social robots for LMICs. The study presents design recommendations for developing social robots that are contextually relevant and aligned with the values and care needs of older adults in Pakistan. Additionally, it compares the cultural dynamics of LMICs with the literature from high-income countries (HICs). | Sunbul Muneeb Ahmad, Muneeb Imtiaz Ahmad, Carolina Fuentes Toro, Nervo Verdezoto Dias and Katarzyna Stawarz | Late Breaking Work |
| 14:30-14:40 | Social Robots in Education: A Biometrically Driven Approach to Supporting Wellbeing | This pilot study explores the comparative effectiveness of human, robotic, and digital facilitators in guiding meditation sessions, focusing on anxiety reduction and perceived trust among university students. Participants engaged in sessions led by each facilitator type, with physiological stress responses measured via biometric sensors, Photoplethysmography (PPG) for heart rate variability and Galvanic Skin Response (GSR) for skin conductance. Subjective data was gathered through the State-Trait Anxiety Inventory (STAI), the Godspeed Questionnaire Series (GQS), and semi-structured post-session interviews. Results show that while robot-led sessions were associated with lower reported trust, they achieved the greatest reductions in anxiety. These findings are particularly relevant in educational contexts, where access to mental health support is often limited. The study highlights the potential of human-centred social robots, responsive to biometric feedback, to deliver scalable and emotionally adaptive wellbeing interventions. Such systems may offer valuable support during periods of heightened stress, including examinations or life transitions, contributing to the evolving role of therapeutic robotics in education. | Aaron Darmudas, Emilia Sobolewska and Carl Strathearn | Late Breaking Work |