3D Multi-Threaded AI Navigation With Pathfinding and Obstacle Avoidance
Jabari BELGRAVE
Software Development
3D Multi-Threaded AI Navigation With Pathfinding and Obstacle Avoidance
The goal for my thesis was to build an AI navigation system capable of managing multiple agents simultaneously in a 3D world. My focus was on integrating A * pathfinding with velocity-based obstacle avoidance( ORCA) in a multi-threaded environment, allowing agents to navigate terrain and avoid collisions dynamically.
I chose this project because I am passionate about Artificial Intelligence in games and wanted to push the limits of my custom engine’ s ability to handle dynamic AI behavior at scale. I was particularly intrigued by the challenge of integrating multi-threading into pathfinding, which is traditionally performance-intensive at larger scales.
The project took approximately eight months to complete. During this time, I designed and implemented several core systems, including a navmesh generator, terrain generator, multithreaded A * pathfinding, and a velocity-based obstacle avoidance system. Each of these components formed the backbone of the simulation, enabling agents to navigate and avoid each other in real-time 3D space.
This project allowed me to deepen my understanding of AI algorithms, parallel programming, and engine architecture. The system emulates large crowd simulation, which can be used in openworld games.
66 SOFTWARE DEVELOPMENT