⭐ If you find this project useful, please consider starring the repo — it helps others discover it!
My first solo IoT project built using an ESP32 and the Hi-Link LD2410C mmWave radar sensor.
The ESP32 creates its own Wi-Fi Access Point, allowing any nearby phone, tablet, or computer to connect directly to the dashboard without requiring an internet connection.
The dashboard displays live radar information using WebSockets for real-time updates.
This project demonstrates how low‑cost mmWave sensors can enable real‑time human detection for smart environments, robotics, and security systems.
Below are the physical components used in this project.
These photos show the LD2410C mmWave radar sensor and the ESP32 DevKit wiring used for the dashboard setup.
Below is the connection layout between the ESP32 and the Hi-Link LD2410C mmWave radar sensor.
Pin Mapping:
- VCC → 5V
- GND → GND
- TX → GPIO16
- RX → GPIO17
- OUT → Not Connected
Use these connections when setting up your breadboard or PCB.
- 📡 ESP32 Wi-Fi Access Point (192.168.4.1)
- 🌐 Standalone operation (no internet required)
- ⚡ Real-time WebSocket communication
- 👤 Human presence detection
- 🚶 Moving target detection
- 🧍 Stationary target detection
- 📏 Distance measurement up to approximately 6 m
- 📱 Mobile-friendly radar dashboard
- 🎯 Software filtering for stable sensor readings
- 🎨 Custom radar-inspired user interface
This project is currently being documented.
Upcoming documentation includes:
- Wiring diagram ✅
- Hardware photos ✅
- Installation guide ✅
- Required libraries ✅
- LD2410C gate configuration
- Demo GIF ✅
- Project architecture
- Future improvements
⚠️ Note:
The LD2410C provides presence and distance information only.
The current radar interface includes a software‑simulated angle visualization to demonstrate a future multi‑sensor tracking concept. The screenshots above show real sensor output, which does not include angle data.
-
Clone the repository
git clone https://github.com/AbdulKadir337/ESP32-LD2410C-Presence-Detection.git
-
Open the project in VS Code (PlatformIO)
Make sure you have the PlatformIO extension installed.
-
Flash the firmware to your ESP32
Select the correct COM port and click "Upload"
-
Power the ESP32
It will automatically start the Wi-Fi Acces Point:
SSID: UAV-RADAR-NET
Password: 12345678
IP: 192.168.4.1 -
Open the dashboard
Connect to the AP and visit: http://192.168.4.1
-
View live radar data
Presence, movement type, and distance update in real time via WebSockets.
This project uses the following PlatformIO/Arduino libraries:
- ld2410 – communication with the LD2410C mmWave radar
- ArduinoJson – JSON encoding for WebSocket messages
- WebSockets – real‑time WebSocket server for the dashboard
- WiFi – ESP32 Wi‑Fi Access Point
- WebServer – serves the dashboard page
- Arduino – core ESP32 Arduino framework
PlatformIO will automatically install all required external libraries when you open the project.






