Extended Kalman Filter for Mobile Robot Localization

This project involved implementing an Extended Kalman Filter (EKF) to estimate the state of a mobile robot, focusing on its 2D position and orientation (x, y, theta). The goal was to accurately localize the robot using velocity commands and sensor data, accounting for noise and uncertainties in both the dynamic model and sensor measurements. The project was carried out in a ROS2 environment, utilizing a simulated robot with landmark-based localization.

Key Features:
- Dynamic Model Implementation: Modeled the robot's dynamics to predict its next state based on translational and rotational velocities. The dynamic model equations accounted for noise, ensuring realistic simulations.
- Sensor Data Integration: Processed sensor data to determine the range and bearing to landmarks, incorporating noise characteristics to reflect real-world conditions.
- Extended Kalman Filter (EKF): Developed an EKF to fuse the predicted states from the dynamic model with the observed sensor data, refining the robot's pose estimate iteratively.
- Visualization and Testing: Used a custom GUI to visualize the real and estimated robot poses, as well as the landmarks. The GUI helped in debugging and validating the EKF implementation.
- Robust Localization: Ensured the robot's localization accuracy even in the presence of noise, by appropriately handling discontinuities and implementing innovation computation techniques.