Multi-Robot Autonomous Frontier Exploration (ROS 2 + Reinforcement Learning)

Two TurtleBot3 robots cooperatively map an unknown arena: Gazebo simulation, per-robot SLAM, map merging, Nav2 navigation, and a frontier-selection brain that is either hand-designed or learned with PPO. The learned policy maps the benchmark arena faster than the heuristic: it reaches 30 m² 21 % sooner (8/8 paired Gazebo runs, p = 0.008).

Folder	Contents
final-product/	The deployable system: ROS 2 package, RL code, trained model, results, graphs, architecture diagrams, demo videos, presentation. Start with final-product/README.md.
resource-guide/	Curated foundational material (ROS 2, SLAM, Nav2, frontier exploration, reinforcement learning) to learn before working on the code

Quick start:

cd final-product && colcon build --symlink-install && source install/setup.bash

Then follow final-product/README.md §3.