Autonomous Mobile Robotics: Localization, Mapping, and Navigation

This repository contains two robotics projects developed in C++ for the Robotics course at Politecnico di Milano. The projects cover the full autonomous navigation pipeline: from low-level kinematics and GPS coordinate transformations to 2D SLAM (Simultaneous Localization and Mapping) and autonomous path planning using the ROS Navigation Stack.

---

Project 1: Robot Odometry & GPS Localization

Folder: first_project/

Overview

This project focuses on calculating high-precision vehicle odometry and transforming global GPS coordinates into a local reference frame. Using telemetry data (speed and steering angle) from a vehicle running on the Monza circuit, we compute the robot's pose in real-time.

Key Features

• Custom Odometry Node: Implemented a ROS C++ node (odometer) that subscribes to vehicle telemetry (/speedsteer) and computes the state using Runge-Kutta kinematics (Ackermann steering model).
• Coordinate Transformations: Handled the mathematical conversion of dual-GPS coordinates from Latitude/Longitude to ECEF (Earth-Centered, Earth-Fixed), and subsequently to ENU (East-North-Up) local reference frames.
• TF Broadcasting: Published the calculated nav_msgs/Odometry and the dynamic tf transform (odom $\rightarrow$ base_link).
• Telemetry Pipeline: Tracked Monza circuit sector times using Haversine formulas and spatial boundary checking.

Usage

1. Build the workspace: catkin_make
2. Run the odometry node: rosrun first_project odometer
3. Play the provided bag file: rosbag play --clock project.bag

---

Project 2: 2D Mapping & Autonomous Navigation (SLAM)

Folder: second_project/

Overview

This project implements a complete SLAM pipeline and autonomous navigation system for a custom-sized differential drive robot ($0.54m \times ...$). It fuses noisy odometry with front and rear LiDAR scans to build a clean 2D map, and then uses Action Servers to navigate the robot to target goals within a simulation.

Key Features

• LiDAR Sensor Fusion: Merged /scan_front and /scan_back topics into a single 360-degree representation using TF2 coordinate transformations.
• Self-Collision Filtering: Implemented a point-filtering algorithm to dynamically remove LiDAR hits that collided with the robot's own chassis, preventing false obstacles in the map.
• SLAM Pipeline: Integrated the processed laserscans and odometry data with standard ROS mapping packages to generate a reliable 2D occupancy grid of the environment.
• Autonomous Navigation: Configured a realistic robot simulation using Stage and the ROS Navigation Stack. Sent asynchronous navigation goals using ROS Action Servers.

Usage

1. Build the workspace: catkin_make
2. Launch the mapping/navigation pipeline: roslaunch second_project main.launch
3. Play the bag file (with simulated time): rosbag play --clock robotics2.bag
4. Use the RViz interface or Action Client to send 2D Nav Goals.

---

Prerequisites & Dependencies

To run these packages, you need an Ubuntu environment with ROS installed (tested on ROS Noetic).

# Install core ROS navigation and mapping dependencies
sudo apt-get install ros-noetic-navigation
sudo apt-get install ros-noetic-gmapping
sudo apt-get install ros-noetic-stage-ros
sudo apt-get install ros-noetic-tf2-geometry-msgs