ESP32 autonomous toy car

The project contains software for assisting a toy car navigation.

Mission plan

Train a machine learning model that learns to navigate an environment and use it to control the physical device.

Training environment

Allows training and downloading the trained model.

https://sergiuionescu.github.io/esp32-auto-car/sym/sym.html

Exporting the tfjs model to tflite

cd converter
docker compose run tensorflow sh convert.sh

Hardware

Components

• Logic board: 1xESP32 (Plusivo)
• Motor controller: 1xTB6612FNG
• Ultrasound sensor: 3x
• 1xTest board
• 3xUltrasound sensor mounts
• Car Chassis
• 2x5V motors
• MPU9250

Circuit connection

ESP32	TB6612FNG	MPU9250	Motor Left	Motor Right	VL53L0X FM	VL53L0X FR	VL53L0X BR	VL53L0X FL	VL53L0X BL
	VM(9V)
VCC	VCC				VCC	VCC	VCC	VCC	VCC
GND	GND				GND	GND	GND	GND	GND
GND	GND
GND	GND
D18	AIN1
D19	AIN2
D23	PWMA
D27	STBY
D26	BIN1
D25	BIN2
D32	PWMB
D21		SDA			SDA	SDA	SDA	SDA	SDA
D22		SCL			SCL	SCL	SCL	SCL	SCL
3V3		VCC
GND		GND
	A1		A1
	A2		A2
	B1			B1
	B2			B2
D5									XSHUT
D33							XSHUT
D12						XSHUT
D13								XSHUT
D14					XSHUT

Demo video