Thymio Robot Navigation System

Autonomous navigation system for Thymio robot using ArUco marker-based localization, Extended Kalman Filter sensor fusion, and A* path planning with obstacle avoidance.

Overview

This project implements a complete autonomous navigation pipeline for the Thymio mobile robot. The robot navigates from a start position to a goal while avoiding obstacles, using visual localization from overhead camera ArUco markers and wheel odometry fused through an Extended Kalman Filter.

Key Features

• ArUco-based Global Localization: Camera-based pose estimation using ArUco markers
• Extended Kalman Filter: Sensor fusion combining vision and odometry for robust pose estimation
• A Path Planning*: Grid-based path planning with obstacle inflation and cost decay
• Pure Pursuit Path Following: Smooth trajectory tracking with lookahead distance
• ANN Obstacle Avoidance: Neural network-based reactive obstacle avoidance integrated into path following
• Finite State Machine: FSM state management (initialization, path following, goal reached)
• Real-time Visualization: Live view of map, planned path, and actual robot trajectory

System Architecture

Core Components

• main.py - Entry point for the navigation system
• robot_fsm.py - Finite state machine orchestrating robot behavior
• global_navigation.py - ArUco detection, localization, and A* path planning
• global_localization.py - Pose calculation from ArUco markers using pixel position & trilateration
• local_ekf.py - Extended Kalman Filter for sensor fusion
• path_follower.py - Pure pursuit controller with integrated obstacle avoidance
• thymio_helper.py - High-level Thymio interface
• thymio.py - Low-level Aseba protocol implementation
• detect_aruco.py - ArUco marker detection utilities
• math_thymio.py - Mathematical utilities for pose transformations

Supporting Files

• camera_calibrator.py - Camera calibration utility
• test_global_nav.py - Standalone testing for navigation components
• environment.yml - Conda environment specification

Hardware Requirements

• Thymio robot
• Overhead camera (tested with 1920x1080 resolution)
• Printed ArUco markers (5x5 100 tag dictionary):
  • Tags 1-4: Map corner markers
  • Tag 0: Robot marker (mounted on robot)
  • Tag 11: Start position marker
  • Tag 12: Goal position marker
  • Tag 10: Obstacle markers (multiple instances)

Installation

1. Create Conda Environment

conda env create -f environment.yml
conda activate bomr_env

2. Camera Calibration

Generate calibration images:

python camera_calibrator.py

Usage

python main.py