Behavior Simulation for identifying critical scenarios for autonomous vehicle safety


Guaranteeing the safety of autonomous systems is a key step towards certification and market entry of such systems. Analyzing and challenging the popular safety layer approach developed by Mobileye/Intel in the benchmarking and testing framework BARK developed at fortiss aims to identify shortcomings.

Project description

The Responsibility Sensitive Safety model (RSS) is a safety layer concept for autonomous vehicles that aims to ensure safety online by establishing contracts between vehicles. Studies analyzing safety and traffic flow shall be conducted in this project.

Research contribution

Identify critical scenarios for an automated vehicle with the RSS acting as a safety layer in respect to usability (scenario goal not reached in time) or safety (yielding collisions), if possible. We also expect to gather insight on how safety envelope models such as RSS shall be best integrated into behavior planning algorithms.

Project duration

01.10.2020 - 31.12.2020

More information