Mars Explorer

Project: Mars Explorer - Autonomous Rover with Real-Time Command and Control

This project involves the development of an autonomous Mars rover prototype using Arduino and Wi-Fi communication. The rover autonomously navigates using ArUco codes, retrieves items, and travels to specific locations based on real-time commands. This system was designed to be robust in harsh environments like Mars, providing safety assistance for astronauts in emergencies.

Project Description

The aim was to create a fully autonomous rover in a clear plastic ball that navigates by scanning ArUco codes to determine its position. The robot operates within a protective sphere to shield critical components from Mars-like conditions, and can travel to required depots, avoid harzardous locations and avoid moving martian boulders. The system communicates over Wi-Fi, integrating embedded systems, control algorithms, and real-time decision-making.

Motors were directly connected to the wheels via machined inserts, optimised for swift rotation and obstacle avoidance. A key aspect of this project was managing the low center of gravity for stability, while implementing a precise PID control for motor driving using a custom-designed strip-board circuit.

Key Features

Autonomous Navigation using ArUco Codes: The rover utilises an on-board Raspberry Pi 3, in communication with an MQTT broker, in order to recieve signals from a satellite that scans ArUco codes, triangulating its position and calculating optimal path navigation to send back to the rover.

Wi-Fi Communication: The rover receives real-time commands via Wi-Fi from the satellite, ensuring autonomous task execution.

Homography: Used various methods for standardising coordinates into a singular reference frame for consistent and reliable commands.

Optical flow: Estimated angle of attack of a moving green ball on the fly by comparing ball coordinates in 2 consecutive frames to find the vector between them.

PID Control System: Implemented PID for stable motor control, ensuring smooth and precise movement across difficult terrain.

Compact Strip-Board Circuit Design: Electronics were carefully placed on a single layer of strip-board to fit within the spherical robot’s constraints.

Travelling Salesman Problem: Implemented modified versions of this algorithm in order to determine ideal order of depots whilst avoiding environmental hazards.

Technologies Used

The following technologies were utilised for this project:

Python: Used for all image processing software along with MQTT communication, satellite-side.

Simulink (MATLAB): Used to design and simulate the PID controller for motor control.

OpenCV: For implamentation of the ArUco code scanning software due to the variety of in-built computer vision libraries and support.

ArUco Codes: Enabled object detection and navigation by triangulating the robot’s position.

I2C Bus: Used for communication with the accelerometer to provide real-time feedback for motor control.

Project Gallery

These images showcase the completed rover and vision system:

Full robotic hardware system, created using an original Kiwi drive, omnidirectional wheels and a Pi HAT.

Image of satellite view while rover required re-positioning due to high glare preventing ArUco code detection.

Project Demonstration

Footage of fully functional rover in operation with randomised targets (red) and danger zones (blue).

Okak