The complexity of realistic autonomous system scenarios with many parameters results in a curse of dimensionality which makes finding bugs and predicting extents of bug profiles an arduous task through formal methods, and a time consuming process in simulation-based approaches. This paper introduces the Eagle Strategy with Local Search, a nature inspired approach based on eagle strategy to discovering bug profile limits for scenario-based validation of autonomous systems. The approach is modular and extendable in terms of the used strategies, and also scalable to n-dimensions of scenario parameters. The approach is integrated with methods of bug classification and bug profile visualization in high dimensionality. The performance of the approach is demonstrated by extensive simulations of an autonomous vehicle validation scenario.
Eagle Strategy with Local Search for Scenario Based Validation of Autonomous Vehicles
2022-03-07
1835222 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Scenario based audible warnings for autonomous vehicles
Europäisches Patentamt | 2019
|Scenario based audible warnings for autonomous vehicles
Europäisches Patentamt | 2018
|Scenario-Based Validation Approach for Commercial Vehicles
SAE Technical Papers | 2020
|Scenario-Based Validation Approach for Commercial Vehicles
British Library Conference Proceedings | 2020
|Europäisches Patentamt | 2023
|