Report + Presentation

In this practical project, students program autonomous robots to navigate the Duckietown environment using learning-based methods. Working in small teams, they implement and compare imitation learning and reinforcement learning approaches to enable a Duckiebot to follow lanes, avoid obstacles, and reach target destinations. Through hands-on experiments in simulation and on real robots, students gain experience in data collection, model training, and policy evaluation. The project emphasizes the trade-offs between supervised and reward-driven learning and highlights the challenges of transferring learned policies from simulation to the real world.

Students gain a practical understanding of how learning-based methods can be applied to autonomous navigation in real-world settings. They learn to design, implement, and evaluate imitation and reinforcement learning algorithms for robot control, bridging the gap between simulation and physical deployment.

Prerequisites: Previous completion of "Deep Learning", "Computer Vision" or "Self-Driving Cars"