Built as an applied AI experiment to explore how agents can learn control policies inside a driving-style simulation. The project focused on the loop between environment design, reward shaping, training stability and visible learning progress as cars moved from poor early behaviour toward more controlled driving.
Autonomous agents
Domain
RL
Method
Driving simulation
Output
Key details
- Designed a learning environment where progress could be inspected visually.
- Explored reward shaping, agent behaviour and training iteration rather than a static model demo.
- Used the project to connect AI theory with an interactive control problem.