Low-Cost Wearable Solution for Mitigating VR Motion Sickness

Project Summary

This project presents a low-cost wearable leg device leveraging embedded Tiny Machine Learning (TinyML) to mitigate Virtual Reality (VR) motion sickness by enabling standing-based locomotion control.

Main Contributions

• Custom IMU dataset collected from 4 participants.
• Preprocessing using frequency-domain (FFT) and statistical feature extraction.
• Lightweight Deep Neural Network (DNN) trained and optimized for microcontroller deployment.
• End-to-end system achieving ~85% macro F1 score with under 100ms latency.
• Custom SteamVR driver integration with Bluetooth HID communication.

Authors:

• Dhanush Balamurali ([email protected])
• Akash Sivasubramanian ([email protected])

Submission:

• This work is submitted as part of the AMOD Artificial Intelligence Project at Trent University, Peterborough, Canada.
• Submission includes all code, dataset, report, and driver files as required.