Fallback Strategy for Level 4+ Automated Driving System

Jing Yu1, Feng Luo1

  • 1Tongji University

Details

Category

Special Session

Sessions

11:00 - 12:00 | Mon 28 Oct | Gallery Room 4 | MoC-T7

Special Session on Solving the Automated Vehicle Safety Assurance Challenge (I)

Full Text

Abstract

According to SAE levels of automation, for systems below Level 3, the fallback roles are human drivers, therefore the automated driving system (ADS) can assume that, when an error occurs, human driver is reliable and will take over the dynamic driving task (DDT) correctly. But for high automation (Level 4) and full automation (Level 5), the fallback roles are no longer human drivers but ADS, which means there will be no driver taking over control when subsystem faults happen. Therefore, it is necessary to define a fallback strategy for level 4+ ADS which can help ADS to bring vehicle to minimal risk condition when failure occurs. In this paper, we propose a fallback strategy structure for Level 4+, which contains three degraded levels with totally 7 different fallback scenarios. When different functional failures occur, ADS is switched to corresponding degraded mode and fallback scenario to bring vehicle to minimal risk condition. The fallback strategy model is established as a state machine in Matlab/Simulink. 6-degree polynomial is applied to output fallback scenarios’ desired trajectory to motion control module, which is designed based on sliding mode control and preview model. The proposed fallback strategy is simulated in Carsim/Simulink environment, whose feasibility of bringing the faulty vehicle to the minimal risk condition is proven.

Additional Information

No information added

Video

No videos found