Autonomous Exploration using 2D LiDAR with Dubins-RRT* Planning and Sampling-Based Goal Selection

This ROS package implements an autonomous exploration system using RRT* with Dubins path constraints and a Pure Pursuit controller for a mobile robot equipped with a 2D LiDAR. It is tested in the Stonefish simulator and supports real-time exploration, goal sampling, path planning, and trajectory execution.

---

Objectives

• Perform autonomous exploration in unknown environments.
• Sample informative goals using information gain.
• Plan feasible, smooth paths using RRT* and Dubins curves.
• Track paths using a Pure Pursuit controller.
• Integrate with simulated or real TurtleBot via ROS.

---

System Architecture

The diagram below shows the data flow between the main components of the exploration system.

How to Run

1. Launch Simulation and All Nodes

roslaunch lidar_based_exploration stonefish.launch

This launches:

• TurtleBot robot model (via turtlebot_basic.launch)
• Octomap server and point cloud conversion
• Online planner and sampling-based goal selector
• RQT GUI and RViz visualization

---

2. Node Overview

Node	Description
laser_scan_to_point_cloud_node.py	Converts /rplidar scan to a usable point cloud
sampling_based_exploration_node.py	Samples goals based on information gain
turtlebot_online_path_planning_node.py	Plans with RRT* and executes with Pure Pursuit
octomap_server_node	Builds occupancy map from LiDAR
rqt_gui	GUI interface

Authors

Gebrecherkos Gebrezgabhier
Pravin Oli
Supervised by: Narcis Palomeras, Sebastian, Martha
University of Girona – Intelligent Field Robotics Systems