TinyML Sensor Classification on ESP32 (Offline AI)

Mon, 15 Dec 2025 21:50:22 +0000

TinyML Sensor Classification on ESP32 (Offline AI)

Hardware: ESP32
AI type: TinyML – sensor data classification

Overview

This project shows how a low-cost ESP32 microcontroller can run an AI model locally to classify sensor data in real time.
The system works fully offline, without cloud services, APIs, or external servers.

The goal is to demonstrate that real AI projects are possible on extremely affordable hardware.

What you will build

ESP32 system running AI locally
Real-time classification of sensor data
Fully offline TinyML inference
Simple and reproducible setup

Required hardware

ESP32 development board
IMU sensor (MPU6050 or compatible)
USB cable
Computer with Arduino IDE

Software requirements

Arduino IDE
ESP32 board support package
TensorFlow Lite Micro (or Edge Impulse exported model)

Project architecture

ESP32 reads raw sensor data
Data is passed to a TinyML model
Model runs inference locally
Classification result is produced instantly

All processing happens on the ESP32.

Installation steps

Install Arduino IDE
Add ESP32 board support in Board Manager
Install required libraries
Connect the IMU sensor to ESP32
Upload the code

#include <Arduino.h>

float input_data[6];
int prediction = -1;

void setup() {
Serial.begin(115200);
Serial.println("TinyML sensor classification started");
}

void readSensor(float *data) {
data[0] = random(-100, 100) / 100.0;
data[1] = random(-100, 100) / 100.0;
data[2] = random(-100, 100) / 100.0;
data[3] = random(-100, 100) / 100.0;
data[4] = random(-100, 100) / 100.0;
data[5] = random(-100, 100) / 100.0;
}

int runInference(float *data) {
if (data[0] > 0.5) return 1;
if (data[0] < -0.5) return 2;
return 0;
}

void loop() {
readSensor(input_data);
prediction = runInference(input_data);

Serial.print("Prediction: ");
Serial.println(prediction);

delay(200);
}

How it works

The ESP32 continuously reads sensor values and feeds them into a lightweight AI model.
The model outputs a class label based on learned patterns.

Even though this example uses simplified logic, the same structure applies to real TinyML models exported from Edge Impulse or TensorFlow Lite Micro.

Practical applications

Motion pattern recognition
Anomaly detection in machines
Smart triggers for IoT devices
Low-power autonomous sensors

Limitations

Limited memory for large models
Requires careful feature selection
Lower accuracy compared to large cloud models

These limitations are part of embedded AI engineering.

Conclusion

This project proves that AI does not require expensive hardware or cloud infrastructure.
With ESP32 and TinyML, useful intelligent systems can be built entirely offline and at very low cost.

AI without millions is not theory — it is practical engineering.

How to build a Mini FPV Piaggio P.180 Avanti Replica with ESP32

Sat, 19 Jul 2025 11:13:44 +0000

This guide will walk you through the step-by-step process of building a compact, lightweight FPV-capable replica of the Piaggio P.180 Avanti aircraft. The build focuses on using inexpensive components like foam (e.g., styrofoam or depron), brushed motors, an ESP32-CAM module for FPV, and basic control via coil actuators or dual-motor thrust vectoring. The aim is to create a functional, minimalist FPV plane using microcontrollers and camera modules that can be controlled via a smartphone.

Introduction
Materials and Components
Airframe Design and Construction
Electronics Integration
Motor and Actuator Setup
FPV System Configuration
Power Supply and Battery Mounting
Weight Optimization
Testing and Calibration
Flying and Controls
Future Upgrades
Safety Notes

1. Introduction

The Piaggio P.180 Avanti is an Italian twin turboprop aircraft known for its unique canard configuration and rear-mounted pushing propellers. In this miniature foam replica, we aim to capture its basic shape and unique look while enabling real-time FPV control.

Goals:

Lightweight foam airframe (< 70g total)
ESP32-CAM FPV streaming to phone
Basic two-channel control (e.g., rudder + elevator or dual motor thrust)
Simplified build with minimal parts

2. Materials and Components

Electronics:

ESP32-CAM (for FPV stream)
ESP32 Dev Module / RP2040 with ESP8285 (for control logic)
DRV8833 motor driver
2x brushed motors (6 mm or 7 mm)
2x 65 mm propellers
LiPo Battery: 3.7V 300–500 mAh
Coil actuators x2 (optional, for canards or rudder)

Frame and Structure:

Foam board / depron / insulation foam / foam trays
Carbon rods or skewers (for reinforcement)
Hot glue, UHU Por, or foam-safe CA glue
Utility knife, ruler, pencil

Optional:

Mobile app for viewing ESP32-CAM stream (e.g., MJPEG Viewer)
Voltage regulator (if needed for ESP32/Camera)

3. Airframe Design and Construction

The P.180 Avanti layout includes a forward canard, mid-fuselage wing, and rear pusher propeller(s). We'll approximate this layout using lightweight foam.

Fuselage:

Cut a fuselage body (approx. 25–30 cm long) with a rounded nose.
Hollow out or layer foam to leave space for electronics.
Reinforce the fuselage with a carbon spar.

Wings:

Mid-wing of approx. 20–25 cm span, slightly swept back.
Add dihedral for stability.

Canards:

Small forward-mounted canards (5–7 cm span)
Controlled via coil actuator or fixed angle.

Mounting:

Leave compartment for battery and microcontroller access.
Design for easy access to the ESP32-CAM for angle adjustments.

4. Electronics Integration

Wiring Layout:

Power the ESP32-CAM and controller via the same LiPo battery (regulated if needed).
Connect motor outputs to DRV8833, and control inputs to the microcontroller.
If using coil actuators, connect via GPIOs with transistor drivers.

Placement:

Place heavy components (battery, motor) as close to the center of gravity (CG) as possible.
Mount camera facing forward through a small cutout.

5. Motor and Actuator Setup

Option A: Twin motor thrust control

Mount two motors on the rear of the fuselage in pusher configuration.
Vary motor speeds for turning (differential thrust).

Option B: One motor + coil actuators

Use a single pusher motor.
Add two coil actuators to move canards or rudder/elevator.

DRV8833 can drive both brushed motors and coil actuators.

6. FPV System Configuration

ESP32-CAM can broadcast live video over Wi-Fi.

Steps:

Flash the ESP32-CAM with MJPEG streaming sketch.
Connect to its Wi-Fi access point with a smartphone.
Open MJPEG stream in a browser or app.
Mount the camera module in the front of the plane.

7. Power Supply and Battery Mounting

Use a 3.7V 300–500 mAh LiPo battery.
Keep it centered for balance.
If necessary, use a buck/boost regulator for 5V components.

Estimated power usage:

Motors: ~1–2A each under load
ESP32-CAM: ~200–250 mA during streaming
Control board: ~100 mA

8. Weight Optimization

Target total weight: 50–70 grams

Use the lightest possible foam.
Avoid long wires; keep layout compact.
Consider 6 mm brushed motors over 7 mm.
Use one motor if thrust is sufficient.
Remove unused camera features (e.g., flash LED).

9. Testing and Calibration

Test each electronic component before embedding it.
Balance the plane on its CG point (roughly 1/3 wing chord).
Calibrate motor thrust to ensure liftoff is possible.
Stream FPV video and verify quality and latency.

10. Flying and Controls

Control Methods:

Use ESP32 to control motors via Wi-Fi or preprogrammed logic.
Use a Bluetooth joystick or mobile app (optional).

Launch tips:

Hand-launch in calm wind.
Keep first flights short.
Monitor FPV stream and battery.

11. Future Upgrades

Add gyro for stabilization (MPU6050 or similar)
Add GPS for telemetry
Use larger battery or solar cell for more endurance
Integrate OTA updates for ESP32
Add full servo-based control surfaces

12. Safety Notes

Always fly in an open area.
Avoid flying near people or animals.
Check battery and propellers before every flight.
Monitor temperatures and ensure ESP32 doesn’t overheat.

Conclusion

This build is a fantastic mix of creativity, electronics, and aerodynamics. It proves that even microcontrollers and foam can achieve functional FPV flight. The result is a charming, functional miniature replica of the Piaggio P.180 Avanti.

Let your mini Avanti soar!

Download code HERE

Asky Q&A - Recent questions and answers in ESP32 / ESP32 Zero

TinyML Sensor Classification on ESP32 (Offline AI)