Problem & Context

Commercial motion-capture systems such as OptiTrack are accurate but expensive and bound to controlled optical setups. For XR, embodied interaction, and robotics research, we wanted a low-cost wearable alternative that could run in ordinary spaces and survive packet loss, drift, and network noise.

Approach & System Overview

Mesquite MoCap is a full-body suit built from 15 wireless ESP32-C3 IMU nodes worn across the body. Each node streams timestamped inertial data over Wi-Fi to a receiver, which fuses orientations and exports BVH skeletons for visualization and downstream use.

At a high level:

• Hardware: 15 ESP32-C3 IMU modules (accelerometer + gyroscope) mounted on straps.
• Firmware: Arduino-based code for synchronized streaming, buffering, and retry logic.
• Host pipeline: Calibration, sensor fusion, inverse kinematics, and BVH export.
• Visualization: Real-time 3D skeleton in Unity / WebXR.

Robustness & Evaluation

We tuned the sensor-fusion and networking stack to:

• Sustain roughly 32 FPS streaming from all 15 nodes.
• Keep end-to-end latency under 15 ms in typical lab conditions.
• Maintain packet delivery around 99–100% over Wi-Fi.

We benchmarked the system against an OptiTrack setup, observing joint-angle error in a small-degree range on key joints for walking and running sequences.

My Contribution

• Co-developed and debugged the ESP32-C3 firmware, including synchronization and buffering.
• Implemented parts of the sensor-fusion stack and helped tune filters and calibration procedures.
• Worked on the BVH export path and host-side tools used for benchmarking and visualization.
• Helped design experiments and interpret results for a paper now under review.

Outcomes & Next Steps

Mesquite demonstrates that affordable, open-source IMU-based mocap can approach optical quality under real-time constraints. Future work includes integrating additional sensing modalities (e.g., depth cameras, environmental sensors) and using the suit as a platform for embodied AI and human–robot interaction studies.