Expert Perception of Teleoperated Social Exercise Robots

Social robots could help address the growing issue of physical inactivity by inspiring users to engage in interactive exercise. Nevertheless, the practical implementation of social exercise robots poses substantial challenges, particularly in terms of personalizing their activities to individuals. We propose that motion-capture-based teleoperation could serve as a viable solution to address these needs by enabling experts to record custom motions that could later be played back without their real-time involvement. To gather feedback about this idea, we conducted semi-structured interviews with eight exercise-therapy professionals. Our findings indicate that experts' attitudes toward social exercise robots become more positive when considering the prospect of teleoperation to record and customize robot behaviors.


INTRODUCTION
Regular physical exercise is an important pillar of a healthy lifestyle.Despite the overwhelming scientifc consensus on its numerous health benefts, there is a concerning global trend of increasing physical inactivity [13].Adding to this challenge, a comprehensive review has highlighted a signifcant reduction in physical activity levels worldwide due to the impact of the COVID-19 pandemic, ranging from 14% to 72% [2].
In the ongoing quest to encourage and enable individuals to maintain an active lifestyle, the notion of employing autonomous robotic systems as exercise coaches is gaining considerable attention.While socially assistive robots have shown promise in the role of exercise coaches [3,9], their practical application in realworld scenarios poses challenges, particularly in adapting to the dynamic and evolving needs of users [12].Currently, these robots typically rely on pre-programmed, gesture-based interactions, employing demonstrations to instruct users in performing various exercises [8,10,15].However, optimizing the motions and responses of these exercise-coaching robots requires the expertise of exercisetherapy professionals who are intimately familiar with the unique needs and physical therapy objectives of their patients.In essence, the efective implementation of these robotic systems in healthcare and exercise therapy requires close collaboration between technology and human expertise.
Usability is a paramount concern for systems designed for clinicians [6].It is imperative that any technology intended for healthcare settings be easy to use and easy to integrate into existing workfows, ensuring that clinicians can smoothly incorporate robotic exercise coaches into their treatment plans.Motion-capture-based teleoperation appears to be the most promising and efective methodology [5] to achieve this integration.It could enable a clinician to use gesture-based movements to intuitively design robot motions that can be replayed later without the involvement of the expert, thus enabling the quick personalization of such robots.
Simple movements and exercises have been successfully imparted to robots using teleoperation [4,11].Nonetheless, to the best of our knowledge, no research has been conducted to gather professional perspectives on teleoperation within the realm of exerciserelated human-robot interaction.Furthermore, research shows that stakeholder involvement can enable the creation of technology that meets their needs and expectations [14].Šabanović posits a bidirectional relationship between society and technology, emphasizing the imperative of involving stakeholders in the robot design process to foster the development of socially robust technologies [16].We posit that teleoperation enables experts to intuitively design motions for an exercise robot, which could be automated with minimal supervision, thereby reducing the workload on healthcare professionals.Thus, we sought to engage with expert stakeholders to ascertain their interest in such a system and gather insights on any features they would fnd valuable.We also took this opportunity to better understand the challenges that physical therapy experts face and the role that social exercise robots could play as tools to support them.Thus, we sought to answer the following research question: How do therapy experts perceive the idea of social robots facilitating exercises through the use of teleoperation technology?
Since, this study is currently ongoing, the scope of this paper includes only a preliminary analysis of the data acquired so far.The rest of the paper is organized as follows: Section 2 describes our study design in detail.The preliminary results are presented in Section 3, followed by the discussion and conclusion in Sections 4 and 5, respectively.

METHODS
To gain insight into expert perceptions of exercise robots and teleoperation, we conducted three-phase semi-structured individual interviews.The study was approved by our institution's ethical review board, and all participants provided written informed consent.These dialogues were conducted over videoconferencing platforms such as Zoom or WebEx.Experts were compensated with gift cards valued at A C 8 per hour of participation for P1-P5 and A C 12 per hour for others; the mid-study pay increase aimed to better recognize participant expertise.The methodological framework of the study adhered to the guidelines put forth by Adams [1], drawing inspiration from the user study orchestrated by Winkle et al. [18].
Fig. 1 shows the study timeline.Participation lasted between 80 and 190 minutes, depending on how much the expert wanted to share, with a mean of 128 minutes.The study began with succinct introductions that segued into the three distinctive phases delineated in the forthcoming subsections.Participants could take a break at the end of every phase.Each interview involved one participant and two researchers, who noted down detailed observations.Four online questionnaires were utilized to supplement the structured oral interviews and garner more quantitative metrics.The conversations were also recorded, contingent on explicit participant consent, with transcripts set to undergo an inductive thematic analysis.Overall, we hypothesize that the ability to teleoperate a socially assistive exercise robot will boost therapist opinions of that robot.

Phase 1: General Therapy
The primary objectives of this phase were to gain a comprehensive understanding of participants' job responsibilities and build rapport.We also inquired about the key metrics they utilized to evaluate client performance, their social interactions with clients, and the challenges they encountered in their roles.At the end of this section, participants completed Q1, a slightly modifed version of the questionnaire designed by Weiss et al. [17], which we term the Robot Acceptance Survey (Q1).This questionnaire seeks to capture participants' initial perspectives on robots.

Phase 2: Exercise Robots
The primary objectives of phase 2 were to familiarize participants with the concept of social exercise robots and serve as a stepping stone to introduce participants to the possibilities of utilizing social robots as exercise partners or guides.We began the process by posing initial questions to gauge the participant's perceptions and understanding of a such a robot.Following the initial inquiry, we proceeded to show four publicly available videos featuring sample exergames with a recent social exercise robot [3] based on the Rethink Robotics Baxter, a stationary human-friendly robot with two arms and a movable face screen.The intention was to provide practical demonstrations of how such robots could be integrated into exercise routines and promote social interaction.
After viewing the videos, participants completed Q2, a subset of the questions from Q1 tailored to this phase: Q2 assesses their attitudes toward the robot, performance expectancy, cultural context, and social agency [17].Additionally, they were also prompted to express their preferences for specifc features they would like to see in such a robot.The questionnaire included several open-ended questions to gather more comprehensive and general feedback from the participants.Following the survey, this phase culminated with a short session when participants were asked specifc questions to delve deeper into their opinions on exercise robots.

Phase 3: Teleoperation
The third phase of the study was designed to capture the participant's viewpoint on teleoperation within the sphere of socially interactive exercise robots.We provided a succinct explanation of the concept of teleoperation (remotely controlling a robot's movements through one's own movements) and then sought their initial impressions of the same.Subsequently, they were shown three video demonstrations of robot arm teleoperation and one showcasing camera-based teleoperation of a robot head and face.Participants thereafter completed a questionnaire (Q3), in which they self-reported measures regarding their perspectives on teleoperation.This survey comprised inquiries into the features they desire in a teleoperation system, as well as open-ended questions to glean additional insights.This phase ended after a detailed dialogue between the expert and the interviewers similar to phase 2.

Final Debrief
Upon the conclusion of the three phases, the participants completed the same Robot Acceptance Survey (Q4) that they completed at the end of phase 1.The experiment then transitioned into a debriefng stage where participants could ofer any additional feedback, express concerns, or pose any lingering questions they might have.Concurrently, the researchers utilized this opportunity to elucidate further research objectives pertaining to this project, and thanked the experts for their generous contribution of time and insights.

Statistical Analysis
We report the median () and the standard deviation () for responses to Q2 and Q3, without statistical analysis.Due to the small number of participants ( = 8) in this within-subjects study, we conducted a non-parametric Wilcoxon signed rank test on the paired samples of the items on the Robot Acceptance Survey administered before (Q1) and after (Q4) the study.We also report the efect sizes using Cohen's d [7].The signifcance-testing level, , was considered to be 0.05 in all cases.

Participants
We defned an expert as any individual engaged in exercise-centric vocations entailing client interactions.These professions spanned orthopedic practitioners, physiatrists, physiotherapists, occupational therapists, researchers specializing in physical therapy, exercise coaches, and trainers afliated with gyms and ftness centers.To obtain a more comprehensive and global perspective, we recruited English-speaking participants on a worldwide scale.Presently, we have conducted a total of eight interviews ( = 8) with participants (age: range = 32 − 52 years, mean= 42.8, = 7.08; gender: 5 male, 3 female) originating from three countries (India, Great Britain, and the USA).Table 1 shows other demographics of the participants.
The range of robotics expertise among our participants varied considerably.Among them, three individuals ( = 3) were classifed as absolute novices, devoid of any prior robotic interaction.One participant ( = 1) was classifed as a beginner, having engaged with commercial robot toys.A further four participants ( = 4) had experienced interactions with more sophisticated robots, such as Questions posed asked if experts were worried about using the robot, thought using it was a good idea, recommended its use, would use it as a part of therapy for their clients, were interested in learning to use it, considered it a social agent, liked its presence, found it nice to work with and could cooperate with it.Positive questions (higher answers better) are colored green, and the negative question is colored pink.Expert responses are diverse but generally trending positive.
Pepper and NAO.We also asked participants about their experience specifcally with Baxter.One participant had had the opportunity to engage with a Baxter robot ( = 1), whilst another reporting having read about it ( = 1).For the remaining six participants ( = 6), the study served as their initial introduction to Baxter.

RESULTS
For this late-breaking report, we examine only a subset of the study's results.Here, we focus on expert sentiment by evaluating the self-reported metrics provided within their survey responses.This analysis reports expert perspectives on exercise robots, followed by their views on teleoperation, and fnally it delves into any potential alterations in their perceptions of robots due to the study.

Expert Views on Social Exercise Robots
Our exploration begins with an analysis of expert views on social exercise robots, based on the questionnaire (Q2) completed during the second phase of the study.Fig. 2 exhibits a visual summary of expert feedback regarding exercise robots.The expert attitudes toward social exercise robots demonstrate considerable diversity.Their opinions display an expansive range with regard to their level of worry ( = 3.30; = 3.45) and the idea of using an exercise robot ( = 7.10; = 3.62).They are moderately willing to recommend these robots ( = 5.05; = 2.82).There is a similar diversity in readiness to use these robots with their clients ( = 5.00; = 2.36).The variety in views also extends to sentiments about the robot's potential utility in facilitating exercise ( = 8.05; = 3.31), its social agency ( = 8.10; = 3.66), and its presence ( = 8.00; = 3.34).However, the variation in opinions seems less when considering the possibility of learning more about the robot ( = The study yielded a rich array of open-ended responses, ofering deeper insights into individual expert perspectives.For instance, 2 appreciated the physical presence of the robot, while 3 believed it could help reduce client fatigue and boredom.Experts 1 and 1 were charmed by the robot's facial expressions. 1 commended the robot's potential to facilitate lower-limb exercises, even in the absence of legs.Additionally, 1 and 1 found value in the robot's ability to merge physical and cognitive training.5 saw the potential for the robot's use in home-based therapy, particularly in enhancing hand coordination.

Expert Views on Teleoperation
Next, we delve into expert opinions on teleoperation gathered from the questionnaire conducted during phase 3 of the interviews.As seen in Fig. 3, the experts expressed a strong appreciation for the idea of teleoperation ( = 9.05; = 0.59).This fondness is refected by notably high median ratings and minimal variability across all questionnaire items.Worries about the system were minimal ( = 1.30; = 1.61), while enthusiasm for its customization potential was substantial ( = 9.05; = 0.69).Clear interest was expressed in learning how to operate the teleoperation system ( = 8.60; = 1.79), as well as in creating new robot behaviors ( = 8.60; = 1.38) and utilizing the system to teach new exercises ( = 8.65; = 1.34).Experts also recognized the system's potential in creating new exercises ( = 9.05; = 1.50) and enhancing social interaction with their clients ( = 8.95; = 0.858).Overall, the experts agreed that it would be a nice system to work with ( = 8.60; = 1.25) and liked the concept ( = 8.15; = 1.79).The open-ended feedback regarding teleoperation was overwhelmingly positive, with experts 1, 5, and 1 expressing a particular liking for the robot's facial expressions.

Expert Opinions on Robots
Finally, we examined the evolution of the experts' opinions by conducting paired t-tests on the pre-and post-study Robot Acceptance Surveys.These results were analyzed from the viewpoint of technology acceptance and object-centered sociality.Overall, expert perception towards robots grew more positive over the course of the study.Of note, with respect to technology acceptance, experts indicated a signifcantly better ability to cooperate with the robot ( = 0.0, = 0.02, = 0.87).Additionally, after the study they felt people would be signifcantly more impressed if they possessed such a robot ( = 3.0, = 0.04, = 0.74).

DISCUSSION
This study sought to examine expert perspectives on social exercise robots and their teleoperation.The gathered feedback yields crucial insights into the perceived usefulness and appeal of various features associated with an exercise robot and its teleoperation system.
There was notable diversity in expert opinions regarding social exercise robots, as evidenced by high standard deviations (Fig. 2).Nevertheless, a consistent theme emerges in their collective appreciation for teleoperation, supporting our hypothesis that the ability to teleoperate a social robot will boost therapist opinions of that robot.This sentiment was evidenced by both high mean scores and low variability across all teleoperation questionnaire items (Fig. 3), confrming a broad consensus among the experts who participated.The Robot Acceptance Survey results also demonstrate that study participation brought about a more favorable perception of the robot in the domains of cooperation and prestige of robot ownership.While initial perceptions of the exercise robot were mixed, the positive results regarding teleoperation indicate that this feature signifcantly alleviated initial concerns about social robots for exercise.Hence, these preliminary results support the overall potential of social exercise robots that can be teleoperated by clinicians to create and record robot movements.

CONCLUSION
Overall, this study shares expert feedback to shed light on the potential and challenges of social exercise robots.The enthusiasm shown for teleoperation strongly supports the idea of using such an approach to enable clinicians to generate robot gestures for exercise.Nonetheless, this study is still at an early stage.Given the diverse participant demographics, there is a need to gather additional data for a more comprehensive overview of expert perceptions.Moreover, the interview data needs to undergo detailed thematic analysis to grasp the nuances of participant responses.Finally, evaluating the data from phase 1 may reveal key challenges that could be addressed by robots.

Figure 1 :
Figure 1: The experiment timeline, highlighting the three primary phases (general therapy, exercise robots, and teleoperation), the approximate time allocation for each stage, and the administered questionnaires (Q1-Q4).

HRI ' 24 Figure 2 :
Figure2: Expert feedback about exercise robots (Q2).Questions posed asked if experts were worried about using the robot, thought using it was a good idea, recommended its use, would use it as a part of therapy for their clients, were interested in learning to use it, considered it a social agent, liked its presence, found it nice to work with and could cooperate with it.Positive questions (higher answers better) are colored green, and the negative question is colored pink.Expert responses are diverse but generally trending positive.
Figure3: Expert feedback about teleoperation (Q3).Questions posed asked if experts were worried about using teleoperation, thought using it was a good idea, would use it to teach new exercises to clients, would create new behaviors with it, were interested in learning to use it, considered that it could facilitate social interactions with client, liked the ability to customize the behavior of the robot, its presence, found it nice to work with and would like to use it.Positive questions are colored green, and negative questions are colored pink.Expert responses are diverse but generally trending positive.Positive questions are colored green, and the negative question is colored pink.Almost all experts were very positive about teleoperation systems.