A map-based model-driven testing framework for automated driving systems
Scenario-based testing has been the primary evaluation approach to the functional safety of Automated Driving Systems (ADSs). Scenarios can be classified as functional, logical, and concrete. Most works in the literature focus on searching concrete scenarios under limited logical scenarios. How to...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Thesis-Doctor of Philosophy |
| Language: | English |
| Published: |
Nanyang Technological University
2023
|
| Subjects: | |
| Online Access: | https://hdl.handle.net/10356/164257 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Nanyang Technological University |
| Language: | English |
| Summary: | Scenario-based testing has been the primary evaluation approach to the functional safety of Automated Driving Systems (ADSs). Scenarios can be classified as functional, logical, and concrete. Most works in the literature focus on searching concrete scenarios under limited logical scenarios. How to systematically define the search space at the logical level remains challenging. We propose a map-based model-driven framework to search for testing scenarios at both logical and concrete levels by modeling the High Definition (HD) maps on which AVs highly depend. The framework consists of the modeling of roads, junctions, as well as the behaviors of other traffic participants. The framework also consists of an automatic HD map generation method for generating unlimited city-driving HD maps based on scenario feature requirements. Experiments on the Baidu Apollo ADS stack show the effectiveness and efficiency of the proposed testing framework. Results have been published at conferences in the related field. |
|---|