Research

• Topic:
  • Developing statistical methods to assess and demonstrate the reliability of the automotive environment perception in the context of automated driving vehicles.
• Background:
  • The environment perception provided by lidar, radar and camera sensors enables advanced driver assistance systems (ADAS) and different levels of functional safe driving automation.
  • Depending on the level of driving automation, failure of the full sensor set could lead to adverse consequences.
  • Due to the large required test effort, a random environment and the difficulty of generating a reference truth in complex driving situations, the safety validation of the sensors is challenging.
• Tasks:
  • Definition of reliability metrics for the environment perception.
  • Developing learning methods that allow to quantify the effect of possible adverse influences on the sensors (e.g. weather, radiation, spoofing).
  • Developing statistical methods to estimate the effect of environmental conditions on the sensors by means of simulation.
  • Studying which aspects of the sensor reliability can be learned without reference truth by exploiting sensor redundancy.