Computer Vision News - May 2019

33 CARLA Autonomous Driving The project welcomes anyone to join. “Everything is open and free” explains Antonio. They expect a large number of participants and around 200 participants have already registered. What Makes CARLA Significant? CARLA utilizes sensors inside a simulator for conducting research on autonomous driving, including perception, planning, control, forecasting, etc. While many simulators exist, CARLA has some significant features. The project is open, not only in terms of code, but also regarding content such as the 3D models, desk tools, and materials. They aren’t any restrictions on the use of CARLA. Without that, researchers in academia wouldn’t have access to such a quality simulator. CARLA already has a community of users and support of start-ups offering services such as maps and models to support autonomous driving simulation. Behind the CARLA Challenge Vladlen Koltun ’s lab at Intel started CARLA at the end of 2016. At that time, the project required a mix of backgrounds: researchers, engineers, and artists. Antonio Lopez recalls the early stages of the project which involved a collaboration between the Computer Vision Center (which he co-founded) in Barcelona, Vladlen Koltun’s lab, and German Ros , the current Project Manager of CARLA. They took everyone's viewpoints to create the first version of the simulator. This stage, besides including the software and the framework itself, also coincided with a research paper based on the simulator, illustrating how to train different models of AI for driving. After more people joined the team, CARLA received additional sponsorship including from Intel and the Toyota Research Institute as wel l as from Challenge Computer Vision News “ Everything is open and free! ” Felipe Codevilla German Ros AntonioM. Lopez Vladlen Koltun On top: the co-organizers Find out more at: carla.org carlachallenge.org