Automobile makers and academic research institutes have researched advanced driver assistance (ADAS) system and automated driving system in order to realize safe city-driving and comfortable long-range-transport. A level-3 automated driving vehicles for motor way will appear around 2020. Moreover, a level-4 automated driving vehicles for limited area also might be realized at the same time. However, a level-2 or level-3 automated driving in general road is tougher challenge than them. It have to detect traffic participants such as pedestrians and other vehicles and driving environment accurately, understand driving scene deeply and determine complicated driving path. Consequently, it is expected that AI (Artificial Intelligence) technologies are applied to the automated driving system for general roads. They will be used for object detection, scene understanding, path-planning, etc. and provide higher performance than previous algorithms. In the meantime, they have some challenging issues. They are black boxes with huge internal parameters. It is very tough for the automotive makers to guarantee their stability and reliability in all actual driving scenes. So, CAE (Computer Aided Engineering) tools such as HIL and MIL will be needed to validate AI functions. Moreover, reinforcement learning is very effective to train the AI function for path-planning. It learns many driving experiences like human under various driving conditions. However, the automotive makers cannot train the AI function on actual public road because of safety reasons. So, CAE tools realizing various virtual traffic scenes are desired to train the AI function. In this speech, the latest technical trend of ADAS and automated driving systems are introduced and the expectations to CAE tools in their research and development are mentioned.
