Professional Training in VR Boxing

UX Researcher, Engineer

Project Overview

To explore the impact of haptic feedback in VR professional training on athletes' performance, we began by conducting observations and interviews to gain a comprehensive understanding of distance control in combat sports. Based on these insights, we designed tasks using Meta Quest HMD and Vive sensors. Finally, we conducted an user study and processed the data using Python to analyze the results.

My Contributions

Had two observations and an interview with 8 participants.
Conducted a user study to discuss distance control in VR boxing.
Used Python to data-visualize the movement trajectory, calculate the performance, and analyze the result.

Role

UX Researcher

Team

1 UX Researcher (me)
1 Unity Engineer

TImeframe

2024/05-2024/06
2 months

Tools

Python
SPSS
Unity
Meta Quest HMD/Vive sensors

Background

As technology continues to advance, an increasing number of sports technologies have emerged to assist athletes, enhancing their performance. Some of these technologies include VR training, which helps athletes with imagery training. However, most VR training currently only provides visual and audio feedback.

We wonder if adding haptic feedback can enhance athletes' performance in VR boxing training.

In our research, we focus on boxers’ distance control performance.

Formative Study:

Observations and Interviews

Distance control is an important topic in combat sports. A good distance control ability allows the boxer to attack effectively and avoid the opponent’s punch. We conducted two observations and interviews to understand how boxers and coaches train and use distance control during training or the game.

(a) interviewing with a professional coach; (b) an observation of a boxing training

After the observations and interviews, we extracted and proposed findings with thematic analysis.

Findings	Quote Example	Design Consideration
Boxers use their punches to control distance	“(In jab punches) when the arms are straightened, and it solidly punches the opponent, we call it a proper punch. Conversely, arms that are not straightened cannot execute full power, while arms that are straightened but do not punch the opponent result in an ineffective punch.”-A5	Training systems should incorporate haptic feedback to enhance distance control training.
Footsteps are a fundamental technique for combat sports athletes in distance-controlling	“It is easy to control the distance between you and your opponent because you can quickly adjust based on a few punches. However, it’s much harder to dynamically adjust while both you and your opponent are moving.”-C2	We emphasize the importance of incorporating movement into the athlete’s training progress.
Training progress often involves multi-tasking	“While we are training, we do not train only one ability; it usually involves the synergy of each ability in the body part.”-C1	Training programs should not only focus on distance control but also incorporate perceptual-cognitive elements.

Design and Implementation

We designed the task based on the findings and design considerations of the formative study. During the task, we ask the participants to dodge the virtual avatar’s random punch, but at the same time, they have to keep a proper distance between the virtual avatar and themselves.

User flow of the task

To have better motion-tracking, we set up HTC light house for the trackers. The participants’ hands, chest, and the robot’s position will be tracked with the HTC trackers. At the same time. the participants will experience the training progress with a Meta Quest VR.

The combination of (a) HTC Vive sensors and (b) Meta Quest pro head-mounted display.

Key Findings → Next Step

Task duration adjustment
Initially, we set the task duration to 1 second, but this proved too fast, even for professional athletes. After working with two testers to optimize the timing, we determined that 2 seconds per task is more reasonable for performance and accuracy.
Hardware sensor limitations → Reduced tracker usage
During the integration of Meta Quest VR and HTC sensors, we discovered that our computer couldn't handle the 8 trackers we initially planned to use. After testing different configurations, we optimized the signal by reducing the number of HTC sensors to 4, achieving a more stable and reliable setup.

User Study:

Professional Boxers’ Experiment

Based on previous research and studies, allowing users to have haptic feedback makes them have better immersion and enjoyment in VR. However, when it comes to performance in VR training progress, few studies talk about it. In our study, we made an early-stage user study to find the relationship between boxers and VR training with or without haptic feedback.

1. Goal

(a) VR training with physical feedback.; (b) VR training without physical feedback

3. Participants

We recruited 8 professional boxers, including 3 females and 5 males. All participants are right-handed boxers with experience in the National Intercollegiate Athletic Games of Boxing.

The participants will experience two different conditions (with and without physical feedback); simultaneously, we will track their movement data with the Vive trackers.

4. Analyze

We used Python to visualize the movement of the boxer in order to make better observations. Besides, we also caculated the boxers’ total movement, average distance between the opponent, punch trajectory similarity, max distance for further analyze.

Data visualize code written by python.

(a) Data visualization (dark red): virtual opponent’s head, (blue): left hand, (green): right hand, (red): middle point of chest.;
(b) Simulation of the data and relative positions

5. Result

During the user study, we observed that most results did not show significant differences, which was contrary to our expectations. While there was a significant difference in the maximum distance of the punches, we could not conclusively determine that this difference was due to the conditions with or without the encountered-type haptic feedback. This is partly because we did not account for the resistance introduced by the moving robot. The other significant difference appeared in the movement ratio. Participants tended to move more when they did not have encountered-type haptic feedback. We hypothesize that the absence of the sandbag, which made the experience seem more virtual, reduced this pressure, allowing participants to move more freely. As a result, they tended to jump, move around more, and attack the virtual opponent from different angles rather than just from the front.

Key Findings → Future Step

Max punch distance variance → Relationship between punch force and sandbag resistance
‍The data revealed a significant difference in maximum punch distance. Still, it’s unclear whether this is due to sandbag resistance or haptic feedback, which seemed to improve the athletes' distance control. Further testing is needed to clarify this distinction.
Athletes find it entertaining → More complex task design with randomization
Athletes reported that the task felt more like a rhythm game than a training exercise, likely due to the predictable punch patterns and timing. To address this, introducing randomized variables in the task design might make the training experience more dynamic and realistic.