Building an Autonomous Robot with Zero Sensors Onboard

This project demonstrates a fully autonomous robot built with zero onboard sensors.

No ultrasonic sensors.No IR modules.No encoders.No IMU.

Instead of relying on hardware-heavy sensing, the robot uses computer vision powered by Python and OpenCV. An overhead camera tracks an ArUco marker mounted on the robot and calculates its real-time position and orientation.

The ESP32 only receives motor speed commands over WiFi and executes them. All intelligence runs externally.

It’s a clean example of how software can replace hardware complexity.

💡 Why Did I Decide to Make It?

Most beginner robotics projects follow the same path:

Add more sensors

• Add more sensors

Add more wiring

• Add more wiring

Add more calibration

• Add more calibration

Deal with sensor noise and drift

• Deal with sensor noise and drift

I wanted to challenge that approach.

The goal was simple:

Can I build an autonomous robot without adding any onboard sensing hardware?Can I build an autonomous robot without adding any onboard sensing hardware?

This project proves that in controlled environments, external computer vision can completely replace traditional sensor stacks — while reducing cost and complexity.

It’s also a powerful learning platform for:

Vision-based robotics

• Vision-based robotics

Differential drive control

• Differential drive control

WiFi-based command systems

• WiFi-based command systems

Path planning fundamentals

• Path planning fundamentals

How Does It Work?

The system is divided into two main parts:

1️ Vision & Control (Computer Side)

An overhead camera captures the workspace.

• An overhead camera captures the workspace.

OpenCV detects the ArUco marker attached to the robot.

• OpenCV detects the ArUco marker attached to the robot.

The marker provides:

X position

• X position

Y position

• Y position

Orientation angle

• Orientation angle
• The marker provides:X positionY positionOrientation angle

A user draws a path on the screen.

• A user draws a path on the screen.

The system calculates heading error and distance error.

• The system calculates heading error and distance error.

Motor speed crrections are computed in real time.

• Motor speed corrections are computed in real time.

2️⃣ Execution (ESP32 Side)

ESP32 connects to WiFi.

• ESP32 connects to WiFi.

Receives motor speed commands via HTTP.

• Receives motor speed commands via HTTP.

Controls motors using PWM through a TB6612 driver.

• Controls motors using PWM through a TB6612 driver.

Executes differential drive motion.

• Executes differential drive motion.

The robot itself has no awareness — it simply follows commands generated by the vision system.

⚙Hardware Used

ESP32 Dev Board

• ESP32 Dev Board

TB6612FNG Motor Driver

• TB6612FNG Motor Driver

2 DC Gear Motor

• 2 DC Gear Motors

Robot Chassis

• Robot Chassis

Printed ArUco Marker

• Printed ArUco Marker

Overhead Webcam

• Overhead Webcam

Total cost: Budget-friendly (2000 INR or 20$)

🎥 Demonstration

https://lnkd.in/dBbFQCRJ

Example section:

Watch the robot follow a drawn path autonomously

🎯 Key Takeaways

✔ Zero onboard sensors✔ Vision-based localization✔ Low hardware complexity✔ Clean software architecture✔ Easily extendable to multi-robot systems

I AM VIKAS I LOVE ROBOTICS❤️✨🍀