Real-Time Avatar Feedback in Virtual Reality Exercise

Effects of Real-Time Avatar Feedback on Exercise Well-Being and Accuracy: Randomized Crossover Trial in Virtual Reality

The goal of this study is to examine whether real-time feedback provided by a virtual trainer avatar during virtual reality (VR) exercise influences participants' exercise well-being and movement accuracy. The study also aims to compare different types of feedback delivered by the avatar.

The main questions this study aims to answer are:

Does real-time avatar feedback improve exercise well-being during VR-based training?

Does real-time avatar feedback improve movement accuracy during exercise?

Are there differences between no feedback, verbal feedback, and combined verbal-gestural feedback?

Researchers will compare three VR exercise conditions: no feedback, verbal feedback, and combined verbal-gestural feedback, using a randomized crossover design so that each participant experiences all conditions.

Participants will:

Take part in three VR exercise sessions, each lasting approximately 15 minutes

Complete the sessions under different feedback conditions, with the order randomized

Perform guided physical exercises while wearing a VR headset and following a virtual trainer avatar

Provide self-reported ratings of exercise experience after each session

This study is an interventional, randomized crossover trial conducted in adults, with all procedures completed during a single study period.

Study Overview

• Status: Completed
• Conditions: Healthy Participants
• Intervention / Treatment: Behavioral: VR Exercise With No Avatar Feedback; Behavioral: VR Exercise With Verbal Avatar Feedback; Behavioral: VR Exercise With Verbal and Gestural Avatar Feedback

Detailed Description

his study investigates the effects of real-time feedback delivered by a virtual trainer avatar during exercise performed in a virtual reality (VR) environment. The intervention is designed to examine how different modes of avatar feedback influence the exercise experience and movement performance during short, guided training sessions.

The study uses a randomized crossover design, in which each participant completes multiple VR exercise sessions under different feedback conditions. The order of conditions is randomized to control for sequence and learning effects. The feedback conditions differ only in the type of information provided by the avatar and include: no feedback, verbal feedback, and combined verbal-gestural feedback. All other aspects of the VR environment, exercise content, and session duration are kept constant across conditions.

The VR application presents a virtual trainer avatar that demonstrates and guides a standardized set of physical exercises. A custom algorithm processes real-time motion data captured by the VR system, including parameters related to movement pace and range of motion. Based on this analysis, the avatar delivers adaptive feedback during the exercise session according to the assigned condition. In the no-feedback condition, the avatar performs the exercises without providing corrective or motivational input.

Each VR session lasts approximately 15 minutes and is separated from the subsequent session by a fixed 7-day interval to minimize carryover effects. The study is conducted within a single study period, and all procedures are completed after participants finish the final VR session. The intervention does not involve any pharmacological treatment or medical devices regulated for clinical use; VR hardware is used solely as a research tool to deliver the exercise environment and capture movement data.

This study is classified as an interventional behavioral trial and is intended to contribute to research on virtual trainers, human-computer interaction, and technology-supported physical activity by systematically examining the role of real-time avatar feedback in VR-based exercise.

• Study Type: Interventional
• Enrollment (Actual): 28
• Phase: Not Applicable

Contacts and Locations

Study Locations

• Poland
  • Warsaw, Poland, 00-608
    • National Information Processing Institute

Participation Criteria

Eligibility Criteria

• Ages Eligible for Study: Adult
• Accepts Healthy Volunteers: Yes

Description

Inclusion Criteria:

• Adults aged 18-50 years
• Ability to provide informed consent
• Ability to perform light to moderate physical exercise
• No contraindications to participation in virtual reality-based exercise

Exclusion Criteria:

• Age below 18 or above 50 years
• Individuals with a high level of habitual physical activity, defined as engaging in structured physical exercise more than three times per week
• Musculoskeletal, neurological, or cardiovascular conditions that limit safe participation in exercise
• Known susceptibility to severe motion sickness or adverse reactions to virtual reality
• Any condition preventing safe use of a VR headset

Study Plan

How is the study designed?

Design Details

• Primary Purpose: Basic Science
• Allocation: Randomized
• Interventional Model: Crossover Assignment
• Masking: None (Open Label)

Number of Arms

3

Arms and Interventions

Participant Group / Arm	Intervention / Treatment
Experimental: No Feedback; Participants perform a virtual reality exercise session without receiving any real-time feedback from the virtual trainer avatar.	Behavioral: VR Exercise With No Avatar Feedback; Participants perform a standardized exercise session in a virtual reality environment guided by a virtual trainer avatar. The avatar demonstrates the exercises but does not provide any real-time verbal or gestural feedback during the session.
Experimental: Verbal Feedback; Participants perform a virtual reality exercise session while receiving real-time verbal feedback from the virtual trainer avatar.	Behavioral: VR Exercise With Verbal Avatar Feedback; Participants perform a standardized exercise session in a virtual reality environment guided by a virtual trainer avatar. Real-time motion data are analyzed by the VR system, and the avatar provides adaptive verbal feedback during the exercise session.
Experimental: Verbal and Gestural Feedback; Participants perform a virtual reality exercise session while receiving combined real-time verbal and gestural feedback from the virtual trainer avatar.	Behavioral: VR Exercise With Verbal and Gestural Avatar Feedback; Participants perform a standardized exercise session in a virtual reality environment guided by a virtual trainer avatar. Real-time motion data are analyzed by the VR system, and the avatar provides adaptive verbal and gestural feedback during the exercise session.

What is the study measuring?

Primary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Exercise Accuracy Index (EAI) based on motion tracking during VR exercise	Exercise accuracy was quantified using the Exercise Accuracy Index (EAI) derived from real-time motion tracking. The EAI reflects how accurately participants reproduced the virtual trainer's movements in terms of pace and range of motion across fourteen moderate-intensity exercises. For each exercise, the trainer's movement served as the reference pattern. Acceptable variability was defined using the distribution of group performance: pace was accurate if it fell within ±1 SD of the trainer's baseline, and range of motion, recorded in absolute values, was accurate if it did not exceed +1 SD from the trainer's benchmark. This accounts for natural inter-individual differences while identifying meaningful deviations. An exercise was classified as correct only if both criteria were met. The EAI ranges from 0 to 14, where 0 indicates no correct exercises and 14 indicates all exercises performed accurately during the session.	During each VR training session across fourteen moderate-intensity exercises (up to three sessions per participant); not assessed during inter-exercise rest periods or during the final low-intensity cool-down exercise.
Exercise Well-Being	Exercise well-being was assessed as participants' immediate emotional state experienced during the VR exercise session. The measure was based on the short mood assessment tool by Wilhelm and Schoebi (2007) and captures momentary affect rather than stable trait well-being. The scale consists of four bipolar items: discontented-contented, unwell-well, tense-relaxed, and agitated-calm. Each item is rated on a 7-point semantic differential in response to the prompt: "During the training session I have just performed, I felt…". Item scores are averaged to produce a single affective well-being score, with higher values indicating more positive affect during exercise.	Immediately after each VR training session via computer-assisted web-based questionnaire (CAWI), up to three sessions per participant.

Secondary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Affective Attitude Toward Exercise Score	Affective attitude toward exercise was assessed using a four-item semantic differential scale based on Crites et al. (1994). The scale comprised the bipolar items: "not relaxed-extremely relaxed", "not satisfied-extremely satisfied", "not happy-extremely happy", and "uncomfortable-extremely comfortable". Participants responded on a 9-point scale to the prompt: "When I think about exercising, I feel…". Item scores were averaged to yield a single affective attitude score, with higher values indicating a more positive affective evaluation of exercise.	Immediately after each VR training session via computer-assisted web-based questionnaire (CAWI), up to three sessions per participant.
Trainee - Trainer Closeness Score	Perceived closeness to the avatar trainer was assessed using the Closeness subscale of the Coach-Athlete Relationship Questionnaire (CART-Q; Jowett & Ntoumanis, 2004) in its Polish adaptation (PLCART-Q). The subscale consists of four items (e.g., "I trust my trainer", "I respect my trainer"), rated on a 7-point Likert scale (1 = definitely no, 7 = definitely yes). Item scores are averaged to produce a single Closeness score, with higher values indicating stronger perceived emotional bond with the trainer.	Immediately after each VR training session via computer-assisted web-based questionnaire (CAWI), up to three sessions per participant.
Trainee - Trainer Complementarity Score	Perceived complementarity in the trainee - trainer interaction was assessed using the Complementarity subscale of the CART-Q (Jowett & Ntoumanis, 2004; Polish PLCART-Q). The subscale includes four items (e.g., "When I am trained by my trainer, I am ready to do my best", "When I am trained by my trainer, I adopt a friendly stance"), rated on a 7-point Likert scale (1 = definitely no, 7 = definitely yes). Item scores are averaged to yield a single Complementarity score, with higher values indicating stronger perceived reciprocity and coordination with the trainer.	Immediately after each VR training session via computer-assisted web-based questionnaire (CAWI), up to three sessions per participant.
Trainee-Trainer Commitment Score	Perceived commitment in the trainee-trainer relationship was assessed using the Commitment subscale of the CART-Q (Jowett & Ntoumanis, 2004; Polish PLCART-Q). The subscale consists of three items (e.g., "I feel close to my trainer", "I am committed to my trainer"), rated on a 7-point Likert scale (1 = definitely no, 7 = definitely yes). Item scores are averaged to produce a single Commitment score, with higher values indicating stronger perceived attachment and willingness to engage with the trainer.	Immediately after each VR training session via computer-assisted web-based questionnaire (CAWI), up to three sessions per participant.
Perceived Exertion Rating	Perceived exertion was assessed using a 7-point rating scale based on Borg (1982). Participants answered the question: "How exerting did you find the training session you have just completed?". Response options ranged from 1 ("very light") to 7 ("very hard"). The reported outcome is the single exertion rating provided after each session, with higher values indicating greater perceived physical effort.	Immediately after each VR training session via computer-assisted web-based questionnaire (CAWI), up to three sessions per participant.
Perceived Competence Score	Perceived competence during exercise was assessed using three items adapted from the Psychological Need Satisfaction in Exercise Scale (Wilson et al., 2006): "able to complete my personal exercise challenge", "confident in my ability to exercise", and "feel good about my ability to exercise". Participants rated each item on a 7-point Likert scale (1 = false, 7 = true) in response to the prompt: "How did you feel during the training session you have just performed?". Item scores are averaged to yield a single perceived competence score, with higher values indicating greater perceived competence during exercise.	Immediately after each VR training session via computer-assisted web-based questionnaire (CAWI), up to three sessions per participant.
Spatial Presence Score in Virtual Reality	Spatial presence was assessed based on the MEC Spatial Presence Questionnaire (MEC-SPQ; Vorderer et al., 2004). The score was derived from four selected items. Example items include "It seemed as though I actually took part in the action of the presentation" and "It seemed to me that I could do whatever I wanted in the environment of the presentation". Items are rated on a 7-point Likert scale (1 = definitely no, 7 = definitely yes). All items are averaged to produce a single spatial presence score, with higher values indicating stronger perceived presence in the virtual environment.	Immediately after each VR training session via computer-assisted web-based questionnaire (CAWI), up to three sessions per participant.

Collaborators and Investigators

• Sponsor: National Information Processing Institute

Study record dates

Study Major Dates

• Study Start (Actual): May 6, 2024
• Primary Completion (Actual): May 24, 2024
• Study Completion (Actual): May 24, 2024

Study Registration Dates

• First Submitted: January 13, 2026
• First Submitted That Met QC Criteria: January 26, 2026
• First Posted (Actual): February 3, 2026

Study Record Updates

• Last Update Posted (Actual): February 3, 2026
• Last Update Submitted That Met QC Criteria: January 26, 2026
• Last Verified: January 1, 2026

More Information

Terms related to this study

• Keywords: psychological well-being; exercise in virtual reality; exercise motivation; affective responses to exercise; augmented feedback in physical activity; human-agent interaction in exercise
• Additional Relevant MeSH Terms: Behavior; Personal Satisfaction; Psychological Well-Being
• Other Study ID Numbers: 4959.13336.14519

Plan for Individual participant data (IPD)

• Plan to Share Individual Participant Data (IPD)?: YES
• IPD Plan Description: De-identified individual participant data underlying the results reported in the publication, including outcome measures.
• IPD Sharing Time Frame: Data will be available upon reasonable request after publication of the primary results.
• IPD Sharing Access Criteria: Access to de-identified data will be granted to researchers who submit a methodologically sound proposal. Requests will be reviewed by the study investigators and data will be shared in accordance with institutional policies and data protection regulations.
• IPD Sharing Supporting Information Type: STUDY_PROTOCOL

Drug and device information, study documents

• Studies a U.S. FDA-regulated drug product: No
• Studies a U.S. FDA-regulated device product: No