CARLA: An Open Urban Driving Simulator

The paper introduces CARLA, an open-source simulator developed from the ground up to support the development, training, and validation of autonomous urban driving systems. Alongside open-source code and protocols, CARLA provides open digital assets, including urban layouts, buildings, and vehicles, that were created for this purpose and can be used freely, and it supports flexible specification of sensor suites and environmental conditions.

Using CARLA, the authors study the performance of three approaches to autonomous driving: a classic modular pipeline, an end-to-end model trained via imitation learning, and an end-to-end model trained via reinforcement learning. The approaches are evaluated in controlled scenarios of increasing difficulty, with performance examined through metrics provided by the platform. This illustrated CARLA's utility for autonomous driving research and offered a freely available, open environment for reproducible experimentation.