Cameras as sensors

Learn

What you’ll learn

• The range of “seeing” sensors, from simple to advanced
• What each can do and which boards they need

Colour sensors

Modules like the TCS34725 measure the red, green and blue light reflected from a surface close by. Great for sorting coloured blocks or spotting a coloured zone on a competition mat.

Smart camera modules

Pixy2 and HuskyLens contain a camera and their own processor. They do the vision work themselves (colour tracking, line tracking, face or tag recognition) and send simple results over I²C, UART or SPI: “red object at x = 140, y = 90”. An Arduino can use them easily.

Camera boards

• ESP32-CAM: a cheap ESP32 with a camera, good for streaming video over Wi-Fi or simple processing.
• Raspberry Pi + camera: a full Linux computer that can run OpenCV and AI models for real computer vision (see Computer vision basics).

What a camera gives a robot

Task	Simplest suitable sensor
Which colour is this block?	Colour sensor
Follow a coloured ball	Pixy2 / HuskyLens
Read a QR code or tag	HuskyLens, or Pi + OpenCV
Recognise many object types	Pi with an AI model

Tip

Pick the simplest sensor that does the job. A $5 colour sensor beats a camera for “is it red or blue?”.

Examples

Example: sorting blocks by colour

A colour sensor under a chute reads each block. If red is highest, a servo flips a gate left; if blue is highest, right.

if (r > g && r > b) gate.write(30);        // red: left
else if (b > r && b > g) gate.write(150);  // blue: right

Example: following a ball with a smart camera

The camera reports the ball’s x position (0–319). The robot turns to keep it centred:

int error = x - 160;            // how far off centre
int turn = error / 4;           // proportional turning
// left motor = base + turn, right motor = base - turn

Practice

Pick the sensorEasy

Which is the simplest sensor for: telling a black floor from a white one; finding a red ball across a room; telling a red block from a blue one held 1 cm away? An IR line sensor; a smart camera (Pixy2/HuskyLens); a colour sensor.

Centre the targetMedium

A smart camera’s image is 320 pixels wide. The target is at x = 250. Should the robot turn left or right, and by how much (using the example’s formula)? Error = 250 − 160 = 90, so turn right; turn = 90 ÷ 4 ≈ 22.

Plan a sorterHard

Sketch a robot that sorts red, green and blue blocks into three bins. List sensors, actuators and the decision rules.

Quick check

1. What makes Pixy2 and HuskyLens easy to use with an Arduino?

They have built-in processors and send simple resultsThey are cheaper than LEDsThey need no powerThey only work with LEGO

2. Which board can run OpenCV for real computer vision?

Arduino UNORaspberry Pimicro:bitA potentiometer

3. For "is this block red or blue?", the simplest sensor is…

A colour sensorA full camera + AIUltrasonicA gyroscope

Check my answers