Embedded Files
The Idea
The Idea
For my dissertation, I chose to explore A* pathfinding, an area I had been eager to work on for some time. Before starting my Master’s, I also set a goal to take full advantage of the university’s robotic technologies, which would have been difficult to access independently. This led me to combine both interests into a single project by researching pathfinding algorithms for robotic footballers.
Research into robotic football has historically driven significant advancements in robotics, much like how game AI has influenced the broader development of artificial intelligence. By working within this context, I was able to investigate pathfinding in a dynamic, real-time environment that closely mirrors real-world challenges.
Personal Notes:
Personal Notes:
For my dissertation, I researched optimal pathfinding solutions for robotic football players, with a particular focus on A* pathfinding. Through this research, I found that while A* provides highly accurate and precise paths, it is not well-suited to the dynamic nature of a football environment.
The primary limitation of A* in this context is its lack of adaptability. In a constantly changing game state, where player positions and obstacles shift in real time, recalculating paths from scratch becomes inefficient. As a result, alternative algorithms such as D* Lite, which builds upon A*, proved more effective. D* Lite reuses previous path data and updates only the affected portions, making it significantly better suited for dynamic scenarios.
This research is explored in greater depth within my dissertation, which you can read below for a full breakdown of the analysis and findings.
The code was created using Python and the ROS (Robotic Operating System) framework. Gazebo was used as the environment to simulate the robot and the robot itself was the TurtleBot3 Burger (I believe they called him Meaty).
Within this system, I implemented and compared two pathfinding algorithms: A* and D* Lite. These were evaluated in a simulated robotic environment to analyse their performance in dynamic navigation scenarios.
Here is a screenshot from RViz, showing the environment mapped by the TurtleBot. The visualised area represents the football pitch as perceived through the robot’s sensors.
Here is the environment, shown in the Gazebo simulation.
Page updated
Google Sites
Report abuse