Mail Delivery Robot (Bayesian Localization)

This project involved designing a mail delivery system using the TurtleBot 3 Waffle Pi, integrating state estimation, localization, and real-time control. The robot was programmed to navigate a closed-loop path, stopping at preselected locations (nodes) identified by unique colors.

Control System

• A PID controller was implemented to guide the robot along the black line path, using a downward-facing camera for real-time feedback.
• Camera data was converted to HSV format for improved color detection, overcoming initial challenges with RGB interpretation.

Localization

• Bayesian Localization was applied to estimate the robot’s position on a topological map, combining state prediction and measurement updates for accurate positioning.
• Extended Kalman Filter (EKF) techniques were used to enhance precision.

Simulation and Optimization

• MATLAB and Simulink were utilized to model robot behavior before deploying the control and localization algorithms in ROS.
• The system was designed to ensure reliable navigation even under challenging conditions, such as communication delays.

Key Features

• Real-time probability-based localization for high-confidence stops.
• Integration of state prediction and sensor measurements to minimize errors.
• System tuning to balance accuracy and efficiency, with recommendations for faster future performance.

This project demonstrated a robust approach to autonomous robot navigation, showcasing problem-solving in:

• Sensor integration.
• Algorithm design.
• Real-world testing.