Assessment of Technical Requirements for Level 3 and Beyond Automated Driving Systems Based on Naturalistic Driving and Accident Data Analysis

Automated driving systems of SAE Level 3 and beyond allow transferring the driving task and responsibility to the vehicle and its automation systems. A crucial challenge for development and real-world performance is the balance between functionality, availability and safety, as a human driver only needs to be available as a fallback after sufficient lead-time. Consequently, automated driving requires enhanced capabilities of sensors, algorithms and actuators. This paper focuses on improved safety and driving comfort of automated vehicles and upcoming technical requirements compared to driver-only or assisted driving. It uses and adapts the state-of-the-art prospective effectiveness assessment method of ADAS to estimate accident avoidance potentials of automated driving systems. The data sources for this analysis are the Strategic Highway Research Program 2 (SHRP2) and the German In-Depth Accident Database (GIDAS). Exemplary automated driving functionalities for highways are prospectively evaluated and the impact on both traffic safety and driving comfort are presented using crash, near-crash and baseline data. Furthermore, relevant technical requirements for corresponding automated driving systems are derived. For an exemplary use-case, possible impacts on system functionality, availability and safety are presented. Additionally, safety potentials of installing high-performance sensors for automated systems of Level 3 and beyond when driving manually are discussed.


  • English

Media Info

  • Media Type: Digital/other
  • Features: Figures; References;
  • Pagination: 12p
  • Monograph Title: 26th International Technical Conference on the Enhanced Safety of Vehicles (ESV): Enabling a Safer Tomorrow

Subject/Index Terms

Filing Info

  • Accession Number: 01762851
  • Record Type: Publication
  • Report/Paper Numbers: 19-0280
  • Files: TRIS, ATRI, USDOT
  • Created Date: Dec 23 2020 4:02PM