Search Space Optimization for Verification and Validation of Autonomous Vehicles
- Delivery
- Available on this site
- Format
- Price
- Non-members (tax incl.):¥1,100 Members (tax incl.):¥880
- Publication code
- 20219038
- Paper/Info type
- Other International Conferences
- Pages
- 1-6(Total 6 p)
- Date of publication
- Sep 2021
- Publisher
- JSAE
- Language
- English
Detailed Information
Author(E) | 1) Elham Asadi, 2) Michael Orgill, 3) Tim Edwards |
---|---|
Affiliation(E) | 1) HORIBA-MIRA Ltd., 2) HORIBA-MIRA Ltd., 3) HORIBA-MIRA Ltd. |
Abstract(E) | This paper focused on making efficient use of simulation resources in a scenario-based V&V framework for a range of ADAS and ADS. Our proposed method automatically specifies and generates diverse scenarios distributed in the search space, with at least some minimum density, mitigating the risk posed by sparse coverage across space and ensuring no critical test cases are left unexplored. This is realized by building and refining a surrogate model of ADS performance, as determined by the proxy of a novel test oracle. |