CARLA, short for CAR Learning to Act, is an open dataset for developing, training, and validating autonomous driving systems. In addition to open source code and protocols, CARLA also provides open and freely usable digital assets (city layouts, buildings, vehicles, etc.).

CARLA is used to study the performance of three autonomous driving approaches: a classic modular pipeline, an end-to-end model trained via imitation learning, and an end-to-end model trained via reinforcement learning.