Elster, Lukas (2025)
Enhancing Radar Model Validation Methodology for Virtual Validation of Automated Driving Functions.
Technische Universität Darmstadt
doi: 10.26083/tuprints-00028962
Ph.D. Thesis, Primary publication, Publisher's Version
Text
Dissertation_Elster_20241222.pdf Copyright Information: CC BY 4.0 International - Creative Commons, Attribution. Download (62MB) |
Item Type: | Ph.D. Thesis | ||||
---|---|---|---|---|---|
Type of entry: | Primary publication | ||||
Title: | Enhancing Radar Model Validation Methodology for Virtual Validation of Automated Driving Functions | ||||
Language: | English | ||||
Referees: | Peters, Prof. Dr. Steven ; Hein, Prof. Dr. Matthias | ||||
Date: | 21 January 2025 | ||||
Place of Publication: | Darmstadt | ||||
Collation: | XV, 142 Seiten | ||||
Date of oral examination: | 26 November 2024 | ||||
DOI: | 10.26083/tuprints-00028962 | ||||
Abstract: | Automated vehicles and mobility services are increasingly becoming part of everyday road traffic. The safety of these systems is of paramount importance. For this reason, safety validation of automated driving systems is playing an increasingly major role. However, it has been shown that safety can no longer be demonstrated through real test drives, as the complexity of the systems and the resulting scenarios do not allow for an economical implementation. For this reason, simulations are used, but they need to be validated. In addition to other sensors, radars are used as an essential component for environmental detection in automated vehicles. Therefore, the effects and uncertainties of radar sensor validation measurements and their impact on the validation of radar sensor models are the focus of this thesis. For this purpose, requirements for acceptance criteria for radar sensor models are derived from radar measurements and quantified using the further developed double validation metric (DVM) in the thesis. It consists of an estimate of the mean error and the variance in the dispersion of uncertainties. The metric is applied to radar data structures in different abstraction levels. Effects can be represented by dedicated measurement setups, objects and environmental conditions. In the following, a measurement setup is derived to isolate the influence of previously identified effects and effect correlations on the radar cross section (RCS). It is shown that the vehicle models investigated have a large influence on this value and that other factors, such as the height of the mounting position, show little sensitivity on the RCS. In addition to the RCS, a detailed analysis of the reflection centers of the investigated objects is carried out. Reference sensors are used in such dynamic scenarios to transfer the driven trajectories into a simulation environment with minimal additional uncertainties. The presented experiments to quantify the uncertainties of the reference sensors and the transfer to the simulation are qualified with the help of the super reference. The tests confirm the accuracy of the reference sensors used. Finally, a metrics-based validation is performed on an exemplary radar model by recording a validation test with a real sensor. Different levels of abstraction of the radar data are analyzed using the DVM. In particular, the combination of the metric results with a satellite image allows an objective root cause analysis. The newly acquired methods and test designs form a further basis for the standardization of validation tests for radar sensors and their models. |
||||
Alternative Abstract: |
|
||||
Status: | Publisher's Version | ||||
URN: | urn:nbn:de:tuda-tuprints-289623 | ||||
Additional Information: | Weiteres Förderprojekt: 16ME0173 VIVALDI |
||||
Classification DDC: | 600 Technology, medicine, applied sciences > 620 Engineering and machine engineering | ||||
Divisions: | 16 Department of Mechanical Engineering > Institute of Automotive Engineering (FZD) | ||||
TU-Projects: | TÜV Rheinland|19A19004E|SETLevel4to5 Bund/BMWi|19A19002S|VVMethoden |
||||
Date Deposited: | 21 Jan 2025 13:04 | ||||
Last Modified: | 21 Jan 2025 13:05 | ||||
URI: | https://tuprints.ulb.tu-darmstadt.de/id/eprint/28962 | ||||
PPN: | |||||
Export: |
View Item |