• Skip to primary navigation
  • Skip to main content
  • About Us
  • Applications
  • Capabilities
  • Software
  • Research
  • People
  • Contact Us

Optical Bio-Sensing Laboratory

Texas A&M University College of Engineering

Software

Software & firmware

Code that runs our science

Firmware, communication protocols and applications we design and write in house for every system in the lab.

Firmware, APIs and software

We build the whole stack.

Every device we design ships with its own firmware, communication protocol and software, written in house, so data flows from the sensor to the analysis without a black box in the middle.

01 SILICON

Embedded firmware

Drivers and real-time acquisition on the microcontrollers inside our wearables, phantoms and instruments.

nRF52840 / ZephyrSTM32Teensy 4.1SPI / I²C ADCsPMIC
02 PROTOCOLS

APIs and links

Custom BLE services, Wi-Fi and mesh networking, encrypted RF links, framed binary packets with checksums, serial command sets and WebSocket bridges.

BLE GATTWi-FiMesh networkingEncrypted RFBinary framingSerial APIsWebSocketsRTC sync
03 APPLICATIONS

Desktop and touch apps

Instrument control, live visualization, calibration and recording on Windows, macOS, Linux and embedded touchscreens.

.NET MAUIPySide6 / QtMATLABWeb UIInstallers
04 ANALYSIS

DSP, physics and ML

Real-time filters, calibration fits, finite-element mechanics, Monte Carlo tolerance and machine-learning design search.

IIR / waveletsCurve fittingFE mechanicsMonte CarloXGBoost
firmware / obsl_frame.h
// sample packet, streamed every DRDYtypedef struct __attribute__((packed)) {  uint8_t  sync[2];    // A5 5A  uint8_t  ch_mask;    // enabled channels  uint16_t counter;    // gap detection  uint8_t  status[3];  uint8_t  data[3*N];  // 24-bit, MSB first  uint8_t  checksum;} obsl_frame_t;

From register to report

Protocols designed for real data.

Our firmware never blocks on the radio or the cable. Every sample carries a counter and a checksum, so the software knows exactly what arrived, what was dropped and when, at the true crystal-timed sample rate.

  • Robust links. Framed packets, gap detection and graceful fallback when a link drops.
  • Configurable. Gain, rate, bias, lead-off and calibration set from the app, stored on the device.
  • Traceable. Synced clocks, session metadata and raw data kept for post-processing.

Why custom, not consumer

Your data, your algorithms, your terms.

Consumer wearables are built to sell in volume and keep data inside one ecosystem. Research and operational programs need the opposite: open access to the raw signal, algorithms that can be inspected and a data path the sponsor controls. That is why we build the device, the firmware and the software ourselves.

Raw signals
Consumer wearableSummarized scores in a vendor cloud. Waveforms rarely exported.
OBSL platformFull ECG, PPG, pressure and motion waveforms at the sample rate the study needs, available at the device.
Algorithms
Consumer wearableProprietary and closed. Accuracy claims come from the vendor.
OBSL platformDocumented pipelines trained on our own ground-truth data, with provenance that survives review.
Profiles
Consumer wearableOne consumer app for everyone.
OBSL platformMission-specific sensing profiles, thresholds and alerts for aircrew, trainees, medics or patients.
Data path
Consumer wearablePhone to vendor servers, on the vendor’s terms.
OBSL platformEncrypted from the sensor through gateways you control into your platform. No third-party cloud required.
Changes
Consumer wearableWait for the next product cycle.
OBSL platformFirmware, protocol or app changes in days, by the engineers who designed the hardware.
Validation
Consumer wearableTested on the vendor’s own sample, on the vendor’s schedule.
OBSL platformCharacterized on physical phantoms across skin tone, motion and heat, then confirmed in IRB-governed human studies.

Consumer devices still have a place: our phantom platform characterizes rings, watches and patches so programs can see exactly what a commercial sensor reports under heat, motion and different skin tones before they field it.

Built in the lab

Software we wrote for our own research.

Animated recreations of five of our in-house tools. Each one controls real hardware in the lab every week.

OBSL WEARABLE // ADS131M0XBLE | .NET MAUI | WINDOWSDEVICE SCANOBSL patchBATTERY87%~14 h remainingCH0 RADIAL mmHgCH1 DIGITAL mmHgCALIBRATIONLinear fit R² 0.998RECORDING20261003_143210.csv118/76121/78116/75MAP 90 HR 72112/72115/74110/71
Windows desktop app | BLE

OBSL Wearable Device GUI

Connects to our wearable patches over Bluetooth Low Energy, streams and filters two pressure channels live, calibrates them to mmHg on the bench, syncs the device clock and records sessions to CSV for analysis.

.NET MAUIC#WinRT BLEScottPlotCustom GATT service
EEG / EOG / EMG ACQUISITION8 CH | 24-BIT | STM32 + MATLABCONFIGURATIONModeEMGRate1000 SPSGainx12Filter20–450 HzNotch60 HzBiasONLead-offOKStop AcquisitionFS 999.7 SPS 0 droppedCH1CH2CH3CH4CH5CH6TOPOMAP RMSSPECTRUM
Acquisition system | Firmware + app

EEG, EOG and EMG Acquisition System

Our own STM32 firmware drives a 24-bit, 8-channel ADS1299 front end and streams checksummed binary packets; the companion app configures gain, bias and lead-off, filters in real time and maps EEG, EOG and EMG activity.

STM32 firmwareBinary packet protocolMATLABReal-time IIRTopographic mapping
MARK-10 TENSILE ANALYZERPYSIDE6 | ESM303 + M7-200LIVE TESTREADYPRELOADTESTINGSAVEDSTRESS–STRAINstrain (%)YOUNG’S MODULUSE (0–10 %)EXPORTSExcel | PDF reportMethods paragraph
Materials testing | Instrument control

Mark-10 Tensile Analyzer

Drives the Mark-10 test stand over its serial protocol with a hard force limit and watchdog, runs silicone phantom tensile tests and turns them into stress–strain curves, moduli, hyperelastic fits and publication reports.

PythonPySide6Serial instrument controlHyperelastic fittingPDF/Excel reports
FLEXSIMSENSOR DESIGN | PHYSICS + MLFlexure 2D3DDynamicHysteresisToleranceDesign MLSTRAIN FIELD µεT1T2full bridgeSIGNAL CHAINFlexureBridgeADC + PGAFilterserror budget in arterial mmHgDESIGN STUDY (ML)
Design tool | Physics + machine learning

FlexSim Sensor Designer

Designs strain-gauged force and pressure sensors end to end, from flexure mechanics, gauges and bonding through the Wheatstone bridge, ADC and filters, with Monte Carlo tolerance analysis and an ML design-study engine.

PythonFinite-element mechanicsMonte Carloscikit-learn / XGBoost3D OpenGL
HUMAN PHANTOM CONSOLETOUCH | JETSON + TEENSY 4.1LIVEPUMPSETUPHISTSETWAVEFORMHRSYSDIARESPSKIN TONE (MONK)TEMP 34.0 °CMOTIONstillwalkruncyclePAUSEP1 INLETP2 OUTLETTMP102TEENSY LIVE 150 HzP1 mmHgP2 mmHg
Embedded touch console | Firmware + web

Human Phantom Console

Runs our patented benchtop hemodynamic phantom from a touchscreen: waveform, skin tone, temperature and motion control, with Teensy firmware streaming dual inlet and outlet pressure through a WebSocket bridge.

Teensy 4.1 firmwareI²C sensorsPython bridgeWebSocketsTouch web UI

© 2016–2026 Optical Bio-Sensing Laboratory Log in

Texas A&M Engineering Experiment Station Logo
  • College of Engineering
  • Facebook
  • Twitter
  • State of Texas
  • Open Records
  • Risk, Fraud & Misconduct Hotline
  • Statewide Search
  • Site Links & Policies
  • Accommodations
  • Environmental Health, Safety & Security
  • Employment