Exploring user engagement with real-time verbal feedback from an exoskeleton-based virtual exercise coach

Raju Maharjan; Sanjana Mendu; Milton Mariani; Saeed Abdullah; John Paulin Hansen

doi:10.1177/20552076241302652

Exploring user engagement with real-time verbal feedback from an exoskeleton-based virtual exercise coach

Digit Health. 2024 Dec 5:10:20552076241302652. doi: 10.1177/20552076241302652. eCollection 2024 Jan-Dec.

Authors

Raju Maharjan¹, Sanjana Mendu², Milton Mariani³, Saeed Abdullah², John Paulin Hansen³

Affiliations

¹ The University of Oklahoma - Norman Campus, Norman, USA.
² Department of Informatics, Pennsylvania State University, University Park, PA, USA.
³ Technical University of Denmark, Lyngby, Hovedstaden Denmark.

Abstract

Objective: Engaging users during physical exercise is crucial for fostering long-term commitment, however, sustaining that engagement remains a significant challenge. This study explores the design of a voice-enabled exoskeleton-based virtual exercise coach (VEC) that provides real-time verbal feedback to enhance user engagement. The objectives of this study are twofold: (i) to compare user engagement with real-time verbal feedback from both VEC and human exercise coach (HEC) during physical exercise, and (ii) to understand users' perceptions and gather their recommendations for improving future VEC technologies.

Methods: We developed an exoskeleton-based VEC that delivers real-time verbal feedback on users' exercise performance. To evaluate its impact on user engagement, we conducted a lab-based mixed-methods study ( $N = 32$ ) over a period of 6 weeks comparing users' engagement with the VEC and HEC using User Engagement Scale (UES) questionnaire and conducted semi-structured interviews to understand users' perceptions of the VEC.

Results: Participants in this study found the VEC more engaging than the HEC, in terms of focused attention ( $Z = 156.5, p < .001$ ) and perceived usability ( $Z = 32, p < .001$ ). Post-interaction interviews revealed that (i) users found the VEC to be engaging, intuitive, easy to use, and convenient; (ii) users perceived the VEC as a valuable training companion that could help reduce the emotional insecurities often associated with going to the gym; and (iii) users expressed a desire for the VEC to have a personality and embodiment that motivates and supports personalized interactions.

Conclusion: Based on our results, we discuss the challenges and implications for designing future voice-enabled VECs that support engaging physical exercises.

Keywords: Conversational agents; engagement; exoskeleton; natural language interface; real-time feedback; speech recognition; user experience; virtual exercise coach.