Computer Vision News - May 2019

32 Challenge and Workshop - CVPR 2019 Computer Vision News The CARLA Autonomous Driving Challenge aims to achieve driving proficiency based on real traffic situations. Each team needs to maneuver a self-driving car through the same set of predetermined routes. Altogether, the challenge consists of four parallel tracks, which differ in the type of input data provided to agents. Challenge For each track, teams will need to follow specific instructions to reach destination points such as “GO- STRAIGHT, TURN-LEFT, or END” while navigating through intersections and roundabouts. Car, or “agents”, will need to contend with real-life driving scenarios such as:  Merging into a lane  Changing lanes  Navigating intersections  Navigating roundabouts  Following traffic lights and signs  Yielding to pedestrians The testing stage begins on May 20 and the challenge officially ends on June 1. Then CARLA will announce the winning team at the CARLA Autonomous Driving Challenge Workshop , to be held at CVPR 2019 in Long Beach, CA. Antonio Lopez , Associate Professor at the Computer Science Department of the Universitat Autònoma de Barcelona (UAB), works as one of the organizers of the CARLA Autonomous Driving Challenge. He shared with us captivating insight on the latest developments of the challenge. …achieve driving proficiency based on real traffic situations.