Advanced Driver Assistance System (ADASAdvanced Driver Assistance System (ADAS))-enabled vehicles capable of operating at SAE J 3016—level 3 or higher extensively employ deep learningDeep Learning-based systems for realizing ADASAdvanced Driver Assistance System (ADAS) features. Such systems need to be validated with all possible scenarios that could unfold during a trip. In this paper, a novel data-driven methodology to automatically generate relevant driving scenarios based on the operational design domain (ODD) in which the vehicle being driven is proposed; variables that are necessary to evaluate the functionality of an AD/ADASAdvanced Driver Assistance System (ADAS) vehicle are identified and a process to quantify the relevance of each of these variables is proposed using a combination of statistical framework such as CatBoostCatboost, Weight of Evidence (WoE), and Information Value (IV)Weight of Evidence & Information Value. An analytical framework is then developed on the basis of analytical hierarchy process (AHP) using real-world accident scenarios data to automatically generate test scenarios. The framework proposed in this study can be adapted to generate relevant test scenarios given the variables in that ODD.
A Data-Driven Test Scenario Generation Framework for AD/ADAS-Enabled Vehicles
Lect. Notes Electrical Eng.
International Conference on Robotics, Control, Automation and Artificial Intelligence ; 2022 November 24, 2022 - November 26, 2022
2023-11-18
13 pages
Aufsatz/Kapitel (Buch)
Elektronische Ressource
Englisch
Automatic Generation Method of Test Scenario for ADAS Based on Complexity
SAE Technical Papers | 2017
|AV/ADAS Safety-Critical Testing Scenario Generation from Vehicle Crash Data
SAE Technical Papers | 2022
|AV/ADAS Safety-Critical Testing Scenario Generation from Vehicle Crash Data
British Library Conference Proceedings | 2022
|AV/ADAS Safety-Critical Testing Scenario Generation from Vehicle Crash Data
British Library Conference Proceedings | 2022
|Automatic Scenario Generation for Simulation-Based Testing of AD/ADAS
SAE Technical Papers | 2023
|