RealGen: Retrieval Augmented Generation
for Controllable Traffic Scenarios

Conventional methods make the model memorize the data distribution for generating. In contrast, our method employs a retriever to query datasets (including external data obtained after training) and uses a generative model to generate scenarios by integrating the information from the retrieved scenarios.

A vehicle (cyan color) goes straight alone on the road.

A vehicle (cyan color) yelids other vehicles.

Two vehicles drive in opposite directions.

Vehicles start to move in an intersection.

A vehicle yields another vehicle in an intersection.