Holistic Driver Perception—Unaddressed Challenges in Autonomous Driving Today

Vehicle automation is still one key trend in the vehicle industry where nowadays just a few advanced series projects barely reach SAE Level 3, and even less series projects achieve SAE Level 4 (and only in very limited conditions). Modern autonomous driving algorithms focus mainly on situation recognition and automated driving decisions. This essentially includes the planning and implementation of trajectories based on environmental sensor data coming from cameras, LiDAR, radar- or ultrasonic-sensors. Currently, one important aspect of automation which is overlooked is the responsibility to monitor the overall technical condition of the vehicle and the road. Even if all relevant vehicle systems have suitable monitoring functions, there are still malfunctions or critical conditions which may not be covered, which a human driver can detect or even handle intuitively. This is also of great importance when considering functional safety in “controllability” (manageability of a fault). Automated and autonomous vehicles must be able to continuously track the overall condition of the vehicle and the immediate environment to ensure that not only faults but also risks can be detected reliably in time.