CARLA simulator’s Post

Dear CARLA users The CARLA team and community go to great lengths continuously expanding the diversity of the CARLA content library to reflect the diversity of traffic environments, buildings, vehicles and road users in real-world driving environments. While CARLA's assets provide a good range of pedestrians ranging from children to adults with different clothing styles it's important to consider that some pavement/crossing users may be utilizing various modes of personal transport or mobility assistance devices. This is why we would like to bring your attention to an excellent, recent asset contribution from the CARLA community. Marc Teuber, working at itemis under the supervision of Andreas Graf and Hamza Ben Haj Ammar, has contributed a wheelchair user pedestrian asset which is now available in the DEV branch of the CARLA simulator. To add a wheelchair to a pedestrian model, first, check that the 'can_use_wheelchair' attribute of your chosen walker blueprint is true. If so, set the 'use_wheelchair' attribute to True then you may spawn a wheelchair user into your simulation. Update your CARLA code and content and check it out now!

We are delighted to announce the winners of the CARLA CVPR Autonomous Grand Challenge 2024!! The Honorable Runner-up prize is awarded to the Tuebingen AI team, who placed first in the MAP track of the CARLA Leaderboard 2.0 and achieved second place in the SENSORS track. The Outstanding Champion award goes to LLM4AD, who achieved an impressive score of 6.9 in the SENSORS track of the Leaderboard 2.0, 23% ahead of the nearest competitor while also placing second in the MAP track. Additionally, LLM4AD are awarded the Innovation Award, for their innovation in using LLaVa and LLaMA in concert to generate waypoints and select challenging scenarios to focus training efficiently. Congratulations to the winners and to all the competitors who submitted excellent work. For more details about the CVPR Autonomous Grand Challenge go to the website: https://lnkd.in/dvDraz6c Thank you all for participating! We hope to see many of you at the workshop.

Dear CARLA community We are excited to notify you that the Unreal Engine 5 version of CARLA is currently accessible in a designated branch of the CARLA GitHub repository. We encourage you to download and build the repository to start experimenting with the new functionality brought to CARLA by Unreal Engine 5.3 which includes: - Improved scene illumination with Lumen - stunningly realistic reflections and ambient lighting - Improved geometry resolution and performance with Nanite - say goodbye to LODs and welcome geometry which adjusts smoothly and continiously with distance from camera, enabling the use of much more detailed geometry without performance costs This version of CARLA currently includes 1 map, Town 10, and 5 vehicles, all of which have been updated with new materials and textures exploiting UE5's advanced features. Further content from CARLA's content library will be migrated/upgraded in the near future. Check out the UE5 development branch here: https://lnkd.in/dBr7mQqp Note!! - The UE5 version of CARLA is a work in progress. Significant changes will be made to the branch throughout the following year. Some functionality is missing from previous versions of CARLA, while other functionality may be unstable or overhauled. We encourage you to experiment but we strongly discourage any mission-critical work based on this version until an official release. Enjoy!

Check out this amazing documentary detailing the CVC's development of AI for autonomous vehicles operating on rural roads. The project aims towards finding economical and sustainable solutions to public transport in isolated rural areas. CARLA is used to train and test the autonomous agent in simulation within a digital twin of a stunning route through the Catalan Pyrenees. The software is then deployed in real world tests in the Alt Pirineu nature park. https://lnkd.in/d9sxF7Xp

We are excited to share another amazing application of the CARLA simulator from the Instituto de Biomecánica (IBV) in Valencia, Spain, where a team has built a realistic human driving platform driven by the CARLA simulator. The CARLA simulator drives 3 large displays atop a mechanical platform to simulate vehicle movement and a host of sensors to monitor human physiological reactions to driving scenarios. Measurements include eyetracking for creating attention maps along with sensors to measure brain activity and other vital signals of the human driver. The Human Autonomous Vehicle (HAV) helps augment the understanding of the interaction between driver and vehicle, leading to insights for advancement in safety and driver assistance technology. Check out the Instituto de Biomecánica de Valencia's website: https://lnkd.in/dsQFJvqn #adas #simulation #carlasimulator