Exploration Algorithm Comparison and Simulation

Project Overview

This project aimed to study autonomous exploration strategies by testing three different algorithms in a 2D grid simulation and later validating the best-performing one in a Gazebo simulation with TurtleBot3. This project had two following stages:

1. Initial Testing in a 2D Simulation Environment:
   • Implemented three different exploration algorithms:
     • Complete Coverage Path Planning (CPP)
     • Next Best View (NBV)
     • Frontier-Based Exploration
   • Simulated exploration in a grid-based environment using Pygame.
   • Compared algorithms based on coverage efficiency, and speed.

1. Final Testing in a Gazebo Simulation:
   • Selected Frontier-Based Exploration due to its superior performance (40% better in computation and time required).
   • Integrated the algorithm into a ROS2 package.
   • Tested it in Gazebo with TurtleBot3, using Nav2 for navigation.

2D Exploration Testing**

In the first phase, the exploration methods were evaluated in a 50x50 grid-based environment with a virtual robot equipped with:

• A simulated sensor with a limited field-of-view (FOV).
• Obstacle detection to avoid unknown areas.
• Goal selection algorithms for different exploration strategies.

Each algorithm was tested independently:

1️⃣ Complete Coverage Path Planning (CPP)

• Used a lawnmower (snake-like) pattern to cover the entire area.
• Ensured full coverage but was not efficient for unknown environments.

2️⃣ Next Best View (NBV)

• Selected the next goal based on information gain.
• Prioritized moving towards areas that maximized newly discovered space.
• Performed better than CPP but still struggled in bigger spaces due to higher computational need.

3️⃣ Frontier-Based Exploration

• Identified frontier cells (edges between known and unknown space).
• Moved towards the closest frontier to incrementally explore.
• Outperformed other approaches in terms of speed and adaptability.

Gazebo Simulation with TurtleBot3

After determining Frontier-Based Exploration as the best approach, we:

1. Implemented the algorithm in ROS2 and integrated it with Nav2.
2. Ran the TurtleBot3 simulation in Gazebo, with separate terminals managing:
   • The TurtleBot3 simulation.
   • The Nav2 navigation stack.
   • The custom exploration node.
3. The robot autonomously explored the environment, selecting frontiers dynamically and avoiding obstacles.

GitHub Repository

The complete implementation is available at:
🔗 GitHub Repository