About Me

Hi 👋, I’m Joey0609, an Embedded, AI and IoT Engineer and Full-Stack Hardware & Software Developer.

I’m currently working on Deep Learning Based Multi-scenario PDR — combining inertial navigation with deep learning for robust pedestrian positioning across diverse scenarios. I’m also actively learning Deep Learning & Large Language Models.

I’m looking to collaborate on research-oriented projects in embedded systems, sensor fusion, and localization, while seeking graduate study opportunities.

“Coding is my favorite pastime. Nothing beats the sense of accomplishment when my ideas finally work.”

🔬 Research Experience

1. Anchor Free Cooperative Positioning System Fusing UWB & INS

May 2025 – Apr 2026 | 👤 Team Leader

Built a complete multi-node cooperative positioning platform, including sensing hardware, intelligent charging management, Web/desktop visualization software, and localization algorithms. The equipment has passed reliability verification and entered mass production.

Core Tech & Real-World Performance

⚡ Hardware Performance: Foot-mounted IMU nodes deliver stable 200Hz inertial data output, with peak refresh rate of 300Hz, balancing low cost with high sampling performance. A 16-channel intelligent charging station supports simultaneous charging, status monitoring, and batch debugging.
🛡️ Algorithm Optimization: Integrated ZUPT (Zero Velocity Update) algorithm significantly suppresses INS cumulative drift, ensuring stable long-duration walking navigation.
📍 Anchor-Free Localization: Designed an MDS auto-alignment algorithm that relies solely on inter-node ranging and inertial data to reconstruct spatial poses under zero-beacon, zero-initial-coordinate conditions.
🔗 Fusion Positioning: UKF filtering is used for tightly-coupled UWB/INS fusion, achieving centimeter-level positioning accuracy.
🧪 Field Tests:
- Indoor (28m × 15m basketball court): 5 nodes running synchronously, 120s continuous motion, 100-frame sampling — straight lines, turns, loops, and other irregular trajectories are accurately reconstructed.
- Outdoor (1000m long-distance route): Multiple repeated test trajectories show strong consistency, closely matching satellite imagery.
- Compatible with walking cadences of 90/100/120 steps/min; periodic errors are effectively suppressed.

Tech Stack: ESP32-C6, STM32, BMI088, ADXL355, FreeRTOS, UWB/IMU Fusion, Electron, MATLAB

2. Deep Learning Based Multi-scenario PDR

Jan 2026 – Present | 👤 Project Leader

Leading the overall design and development of a deep learning enhanced Pedestrian Dead Reckoning system. Covering hardware adaptation, algorithm research, model deployment and system verification across diverse walking scenarios.

🚀 Featured Projects

NOTAM Website Viewer

🌐 Website: https://joey0609.github.io/notams/

Parses NOTAMs (Notice to Airmen) and maritime safety notices to extract rocket launch areas, debris zones, and restricted zones, then visualizes them on an interactive map.

Key Features:

Automated real-time crawling and parsing of multi-source NOTAM & MSI data
Interactive map visualization with historical archive matching and source reconciliation
Coordinate lookup, distance measurement, and image export
Time window parsing and deduplication across heterogeneous notice formats

Rocket Launch Live Control & Visualization System

joey0609

rocket-mission-control

GitHub repository

———

Real-time rocket launch monitoring and control interface with live telemetry visualization
Web-based command & control system with multi-client synchronization
Integrated data logging and playback for post-launch analysis