Distributed multi-domain simulation and data-driven verification for autonomous driving

One-stop solution service for model exchanging to realize digital transformation in vehicle development

The development of autonomous driving system requires verification using a large number of driving scenarios, and it is essential to utilize advanced simulation technologies. We introduces a distributed simulation framework which combines autonomous driving software and multi domain simulation of vehicle dynamics, vehicle controllers, sensors and road simulation. This framework offers a collaborative virtual workspace across design teams (e.g. OEM and suppliers) over the internet, allowing models and simulations to be interconnected.
We also introduces a data-driven verification methodology which combines both data-to-scenario and knowledge-to-scenario approach. The point of this methodology is to utilize both feature model and behavior model for expressing the driving scenarios and significantly reduces the number of test cases while improving test coverage.
We shows an autonomous driving software verification environment by using distributed simulation platform that combines multiple commercial/open-source simulators, Japanese METI MBD-WG guideline compliant vehicle dynamics and controller model and others.
We also shows a case study of data-driven verification of autonomous driving by using NHTSA (National Highway Traffic Safety Administration) and Japanese SAKURA32 safety methodology.