Movement and Muscle in VR/XR Soon available on the market; details are here
Wearable brain–computer interface (BCI) and EMG/IMU motion capture for VR and VRChat. See YouTube video
Aura VR is an open hardware + firmware + SDK platform built around the STM32WB55 wireless MCU and Texas Instruments ADS1299 24-bit biosignal front‑ends. It streams muscle activity (EMG) and motion (accelerometer + gyroscope) over Bluetooth Low Energy, so a person's real arm and leg movement can drive an avatar in VR / VRChat in real time.
The board can run as a full 16‑channel biosignal recorder (
IronBCI‑16), but for the XR / VR use case you typically start small — 4 EMG channels for the muscles you care about, plus the on‑board IMU for limb orientation.
- Reads EMG from the muscles of the arms and legs with a research‑grade 24‑bit ADS1299 front‑end (the same class of chip used for EEG).
- Reads motion from an on‑board LSM6DS3 6‑axis IMU (3‑axis accelerometer
- 3‑axis gyroscope) to track limb orientation and movement.
- Streams everything over BLE to a host (PC / Raspberry Pi) in a single compact packet.
- Feeds VR: the IMU stream animates a 3D body / limb model; the EMG stream reports muscle contraction (e.g. clenching, flexing, gripping).
- Bridges to VRChat (roadmap): a PiEEG Server forwards movement and muscle activity into VRChat via OSC so an avatar mirrors the wearer's real motion and effort.
Typical applications: embodied VR / VRChat avatars, gesture and grip control, rehabilitation and biofeedback, and general EMG/EEG research.
Two board families share the same firmware base and BLE protocol:
| Board | MCU | Analog front‑end | Channels | Motion | Best for |
|---|---|---|---|---|---|
| Aura VR | STM32WB55 | 1× ADS1299 | 8 EMG/EEG | LSM6DS3 IMU (accel + gyro) | VR / VRChat, EMG + motion |
Additional sensors present on the design: MAX30102 (optical PPG / heart‑rate, optional).
Key specs (ADS1299)
- Resolution: 24‑bit, signed
- Sample rate: 250 SPS (
CONFIG1 = 0x96) - Reference:
VREF = 4.5 V, gain = 1 → 1 LSB ≈ 0.536 µV
IMU (LSM6DS3)
- Accelerometer: ±2 g
- Gyroscope: 245 dps
- Output data rate: 416 Hz (
set_speed = 0x60)
Schematics, Gerber/DipTrace design files, BOMs and the compiled .hex are in the
board folders (ironbci_16/, EMG_VR/).
Advertised name: Aura VR
| Role | UUID |
|---|---|
| Service | 0000fe40-cc7a-482a-984a-7f2ed5b3e58f |
| Notify (data) | 0000fe42-8e22-4541-9d4c-21edae82ed19 |
| Write (command) | 0000fe41-8e22-4541-9d4c-21edae82ed19 |
Streaming starts automatically when the host subscribes to notifications.
byte 0 IMU status
bytes 1..6 accelerometer X, Y, Z (int16, little‑endian)
bytes 7..12 gyroscope X, Y, Z (int16, little‑endian)
bytes 13..36 EMG sample 1 CH1..CH8 (8 ch × 3 bytes, 24‑bit)
bytes 37..60 EMG sample 2 CH1..CH8
bytes 61..84 EMG sample 3 CH1..CH8
bytes 85..108 EMG sample 4 CH1..CH8
EMG/EEG channels are big‑endian signed 24‑bit. Convert one channel to microvolts:
raw = (b0 << 16) | (b1 << 8) | b2 # 3 bytes, MSB first
if raw & 0x800000: # 24-bit two's complement
raw -= 1 << 24
microvolts = 1_000_000 * 4.5 * raw / (2**23 - 1)IMU axes are int16 little‑endian; scale by the configured full‑scale ranges (±2 g accel, 245 dps gyro).
EMG_VR/ Aura VR (EMG + IMU) — the VR path
├── Framework/ STM32 firmware sources (p2p_server*.c)
├── project/EMG_VR.rar Full STM32CubeIDE project (Aura VR)
└── SDK/ Python host software
├── 1_Visualisation_Graph.py Live accelerometer + gyroscope plot
├── 2_body_rotation.py 3D object follows physical IMU rotation
├── 3_body_graph.py IMU plots + 3D object (complementary filter)
├── 4_Body_EMG.py 3D IMU orientation + 8‑ch EMG
├── 4_1_Body_EMG.py 3D IMU + 8‑ch EMG in µV (109‑byte packet)
├── 5.Palm_EMG.py Hand / palm model driven by EMG (hand_fast.obj)
└── hand_fast.obj 3D hand mesh
Host: Windows / Linux / macOS with a BLE adapter (or a Raspberry Pi).
python -m pip install bleak numpy matplotlib
# fast 16-channel plotter also needs:
python -m pip install pyqtgraph PyQt5Power the board; it advertises as Aura VR. No pairing PIN is
required — the host just subscribes to the notify characteristic.
Live IMU (accelerometer + gyroscope):
python EMG_VR/SDK/1_Visualisation_Graph.py3D orientation — the model follows your real movement:
python EMG_VR/SDK/2_body_rotation.py
# press R to reset the current pose to zeroEMG + 3D motion together:
python EMG_VR/SDK/4_1_Body_EMG.py16‑channel recorder (IronBCI‑16):
python ironbci_16/SDK/1.Python.py # scan + plot
python ironbci_16/SDK/1.Python.py --seconds 10 --csv out.csv # log to CSV
python ironbci_16/SDK/2.Python.py # fast pyqtgraph plotConnect by address to skip the name scan:
python EMG_VR/SDK/2_body_rotation.py --address AA:BB:CC:DD:EE:FFThe XR pipeline turns two physical signals into avatar behaviour:
- Motion → limb orientation. The IMU (accelerometer + gyroscope) is fused (complementary filter → pitch / roll / yaw) and mapped onto a 3D limb so the avatar's arm or leg follows the real one.
- EMG → muscle activity. The rectified/smoothed EMG envelope reports how hard a muscle is contracting — useful for grip, flex, and gesture triggers on arms and legs.
The intended VRChat path is a lightweight PiEEG Server that:
- Connects to the board over BLE and decodes the packet above.
- Converts IMU → joint rotations and EMG → normalized muscle‑activation values.
- Sends them to VRChat over OSC (
/avatar/parameters/...) so an avatar mirrors the wearer's real movement and muscle effort in real time.
Status: the BLE acquisition and 3D/EMG visualization are implemented in the
SDK/scripts. The OSC bridge to VRChat is the next milestone and is not yet included in this repository.
You don't need all 16 (or even all 8) channels to get moving. A common starting configuration is 4 EMG channels on the muscles you're targeting (for example forearm flexor/extensor for grip, plus biceps or a leg muscle), together with the IMU for limb orientation. Read the full packet as usual and simply use the subset of channels you've wired up; unused channels can be left unconnected or ignored in the host script.
- BLE acquisition (8‑ch EMG + IMU, and 16‑ch EEG/EMG)
- Live IMU plots and 3D orientation viewer
- EMG visualization in microvolts + hand/palm model
- PiEEG Server: OSC bridge to VRChat (movement + muscle activity → avatar)
- Configurable channel selection / gain from the host
- Calibration and per‑muscle activation mapping presets
This is research and hobbyist hardware, not a medical device. Use only with appropriate, isolated power when placing electrodes on the body, and do not use it for diagnosis or treatment.
Built on the PiEEG / IronBCI biosignal platform (STM32WB55 + ADS1299 + LSM6DS3). See the board folders for schematics, BOMs and firmware.