The development and deployment of autonomous driving (AD) systems are very challenging from a safety perspective. In addition to the main AD processing pipeline, self-driving cars need a safety assessment system that can evaluate the degree of safety of the ego vehicle with respect to its environment at run time. As early risk detection is very important in case of a dangerous situation or a system failure, the safety assessment is required be “predictive”, i.e. the evaluated degree of safety needs to not only consider the current state but also anticipate the dynamic evolution of the road traffic. In this work, we present a predictive safety assessment system for level 3 autonomous driving by assuming certain motion models for the ego vehicle and the other road users. Firstly we formulate both longitudinal and lateral safety as constrained optimization problems. Then we propose dynamic programming (DP) algorithms to solve these problems in real time. The DP algorithm for longitudinal safety solves the problem exactly by taking advantage of the special properties of the problem, whereas the DP algorithm for lateral safety solves the problem approximately by employing efficient approximate control policies. Experimental results on simulated driving scenarios demonstrated the effectiveness of the proposed safety assessment system.
Real-Time Model Predictive Safety Assessment for Level 3 Autonomous Driving
Lect. Notes in Networks, Syst.
International Conference on Robot Intelligence Technology and Applications ; 2021 ; Daejeon, Korea (Republic of) December 16, 2021 - December 17, 2021
01.04.2022
13 pages
Aufsatz/Kapitel (Buch)
Elektronische Ressource
Englisch
Real-Time Implementation of Randomized Model Predictive Control for Autonomous Driving
IEEE | 2022
|Model Predictive Control for Safe Autonomous Driving Applications
Springer Verlag | 2022
|Real-time autonomous driving sensor calibration apparatus and method during autonomous driving
Europäisches Patentamt | 2023
|