Study on the Influence of Embodied Avatars on Gait Parameters in Virtual Environments and Real World

Tianyi Zhou, Ding Ding, Shengyu Wang, Chuhan Shi, Xiangyu Xu

TL;DR

This study reveals that embodying a look-alike "old-age" avatar in virtual reality (VR) significantly alters a person's gait, not just within the simulation, but in the real world immediately afterward. The researchers found that while VR walking is generally slower and more cautious than real-world walking, the "Proteus effect"—where users adopt the characteristics of their digital avatars—creates a measurable carry-over effect on physical movement.

What Problem Does This Paper Address?

Virtual Reality is increasingly used for clinical gait analysis and rehabilitation for conditions like Parkinson’s disease. However, two major questions remain:

  1. Ecological Validity: Does the way we walk in a headset actually reflect how we walk in the real world?
  2. The Proteus Effect: Does the appearance of our digital self (our avatar) subconsciously change our physical behavior?

Most previous studies used "incomplete" avatars (e.g., just floating hands or feet). This paper investigates whether a complete, look-alike avatar—specifically one that represents the user at a much older age—influences gait parameters like stride length and velocity in both virtual and physical environments.

Methods

The researchers developed a custom VR system using Unity3D and an HTC VIVE Pro. To ensure high realism, they used LiDAR technology to create a "digital twin" of a real-world 5-meter corridor.

Key technical components included:

  • Avatar Generation: Using Character Creator 4, the team generated two look-alike avatars for each participant based on their photos: a "same-age" version and an "old-age" version.
  • Motion Capture: An Azure Kinect sensor tracked 10 vital joint points of the lower body in real-time.
  • Gait Analysis: The team used paired-sample t-tests for statistical analysis and implemented a hybrid CNN-LSTM-Attention neural network. This model was designed to classify gait patterns and determine if the differences between environments were distinct enough to be recognized by an AI.
Fig. 1: Avatar created from a user’s photo: (a) the same-age avatar, and (b) the old-age avatar. (c) The avatar the user sees in the VR scene.

Datasets

The study involved an original dataset collected from 26 healthy college students (aged 18–26). Participants were split into groups to test different sequences of walking in the real world versus walking in VR with different avatars. No external benchmarks were used, as the study focused on the internal comparison of "Real World" vs. "Virtual Environment" gait for the same individuals.

Key Insights and Ideas

  • The "Cautious" Virtual Gait: Even with a look-alike avatar, participants walked differently in VR. Stride length, velocity, and cadence were all significantly lower in the virtual environment compared to the real corridor. This suggests that "VR gait" is a distinct behavioral pattern, likely due to a lack of physical feedback or subconscious caution.
  • The Carry-Over Effect: The most significant finding is that the Proteus effect isn't confined to the headset. After embodying an elderly avatar in VR, participants' real-world walking velocity and step length decreased. This suggests that the brain "retains" the digital persona's characteristics for a period after the simulation ends.
  • Neural Networks as Discriminators: The researchers used their deep learning model not just to analyze data, but as a proof of difference. The model could distinguish between real-world and VR gait with 91% accuracy, confirming that the two modes of walking are fundamentally different.
Fig. 3: The corresponding names of the captured joint points

Key Results

  • Environment Impact: In VR, stride length dropped from an average of 1.241m to 1.089m, and velocity dropped from 1.214 m/s to 0.856 m/s ($p < 0.001$).
  • Avatar Impact (VR): When wearing the old-age avatar, participants' step width was significantly lower than when wearing the same-age avatar ($p = 0.027$).
  • Post-VR Impact (Real World): After the old-age VR experience, participants' real-world velocity dropped from 0.929 m/s to 0.895 m/s ($p = 0.042$), and step length decreased from 0.529m to 0.516m ($p = 0.030$).
  • Subjective Experience: Questionnaires confirmed that participants felt a strong sense of "ownership" and "agency" over their avatars, which likely fueled the behavioral changes.

Limitations and Future Work

The study's primary limitation is its demographic: the participants were all healthy young university students. It remains to be seen if these effects are more or less pronounced in older adults or patients with existing gait impairments.

Additionally, the "old-age" avatar only changed the participant's appearance. Future work could involve simulating physical age-related constraints, such as a "hunchback" posture or reduced joint flexibility, to see if these further intensify the Proteus effect. The authors also suggest using multiple Kinect cameras or wearable IMU sensors (gyroscopes/accelerometers) in future iterations to capture more nuanced 3D gait features.

Summary updated

Abstract

In this study, we compare the virtual and real gait parameters to investigate the effect of appearances of embodied avatars and virtual reality experience on gait in physical and virtual environments. We developed a virtual environment simulation and gait detection system for analyzing gait. The system transfers real-life scenarios into a realistic presentation in the virtual environment and provides look-alike same-age and old-age avatars for participants. We conducted an empirical study and used subjective questionnaires to evaluate participants' feelings about the virtual reality experience. Also, the paired sample t-test and neural network were implemented to analyze gait differences. The results suggest that there are disparities in gait between virtual and real environments. Also, the appearance of embodied avatars could influence the gait parameters in the virtual environment. Moreover, the experience of embodying old-age avatars affects the gait in the real world.
BibTeX
@misc{zhou2024study,
  author = {Tianyi Zhou and Ding Ding and Shengyu Wang and Chuhan Shi and Xiangyu Xu},
  title = {{Study on the Influence of Embodied Avatars on Gait Parameters in Virtual Environments and Real World}},
  year = {2024},
  eprint = {2411.18949},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2411.18949},
}
Figure