Urban Driving with a Model-Based Reinforcement Learning Agent
A Dreamer-inspired agent that learns to drive in a simulator and keeps adapting to the driver for a more comfortable ride.
A model-based reinforcement learning agent, inspired by DreamerV2, that learns to drive in urban areas using a simulator.
On top of the agent, I designed and implemented a continual learning framework, so the system keeps improving over time and adapts its behaviour to the driver’s, making the ride more comfortable.