Valider la conduite des véhicules autonomes grâce à des milliards de kilomètres d'essais virtuels.
"Virtual testing is of the most vital importance to autonomous vehicle development, since the autonomous system itself needs to run through billions of miles of virtual road testing or real road testing to ensure SAE level 5 safety requirements. As road network simulation is an essential part of any virtual testing environment, the ability to create and share billions of miles of road network data becomes ever more critical.
This paper introduces the rationale behind the Open Standards movement for virtual road simulations, and why standardization is essential to the mobility industry’s collaboration on developing safe, fully automated, self-driving vehicles.
Road Network Data - The Bottleneck of Autonomous Driving Development
Several commercial and open-source vehicle simulation environments are currently in use in order to train the perception and path-planning of autonomous driving algorithms. What they all share in common, is the need for High Definition Maps (HD Maps) of road networks. Autonomous developers and engineers care about HD Maps because these maps are extremely precise (at centimeter-level for road features) and provide key navigational input to autonomous vehicles. This precision is necessary so that the systems can make real-time decisions based on complex road situations like urban downtown areas. As such, HD Maps become a crucial element of the development process of autonomous vehicles."
