Is this project an undergraduate, graduate, or faculty project?
Graduate
individual
What campus are you from?
Daytona Beach
Authors' Class Standing
PhD Student
Lead Presenter's Name
Quentin Goss
Faculty Mentor Name
Dr. Mustafa Ilhan Akbas
Abstract
Today is an age of exiting emerging technology where cutting-edge research in autonomous vehicles (AVs) reduces the active human participation in driving and extends awareness beyond human limitations of perception and reaction, improving driving safety and quality of the user experience as a result. The ever-increasing complexity of these autonomous systems poses many challenges towards the validation and verification (V\&V) of these complex systems under time and resource constraints, as the use of artificial intelligence and also the intricacy of the operating environment means that these systems are also black-box and non-deterministic. Scenario-based V\&V testing of such systems, which involves constructing relevant and realistic scenarios to evaluate the performance of a complex system through observation, is a practical and effective means of V\&V testing under time and resource constraints. Furthermore, scenario-based V\&V testing has been the hallmark approach when testing the AV's predecessor, the automobile, as such it is the natural progression to perpetuate scenario-based V\&V testing for AVs. As such, this dissertation proposes an end-to-end scenario-based V\&V testing framework for AV V\&V testing, which intelligently and effectively identifies and proposes solutions to the challenges of testing AVs and other complex systems under time and resource constraints. The dissertation begins by identifying sources of AV V\&V scenarios, then constructing modular and measurable scenarios that are validated through scenario-based tests. Then, challenges in scenario-based testing of highly complex, high-dimensional scenarios are identified through modeling. This dissertation includes numerous studies of sequence sampling and exploration strategies to efficiently explore high-dimensional spaces, and to select high-value and edge scenarios. After the foundations of the proposed framework are validated, the dissertation diversifies: a first-of-it's kind AV scenario-description language for network-based traffic simulation is proposed, a practical means of estimating scenario complexity of AV V\&V scenarios is proposed, and interpretation of the scenario test data by leveraging explainable artificial intelligence is proposed. Finally, the framework is stress tested via highly complex, high-dimensional scenario tests to evaluate the scalability, interpretability, and performance of the dissertation research.
Did this research project receive funding support from the Office of Undergraduate Research.
No
ScenarioXP: A Complete Scenario–Based Testing Framework for the Exploration and Exploitation of Autonomous Vehicle Validation Scenarios
Today is an age of exiting emerging technology where cutting-edge research in autonomous vehicles (AVs) reduces the active human participation in driving and extends awareness beyond human limitations of perception and reaction, improving driving safety and quality of the user experience as a result. The ever-increasing complexity of these autonomous systems poses many challenges towards the validation and verification (V\&V) of these complex systems under time and resource constraints, as the use of artificial intelligence and also the intricacy of the operating environment means that these systems are also black-box and non-deterministic. Scenario-based V\&V testing of such systems, which involves constructing relevant and realistic scenarios to evaluate the performance of a complex system through observation, is a practical and effective means of V\&V testing under time and resource constraints. Furthermore, scenario-based V\&V testing has been the hallmark approach when testing the AV's predecessor, the automobile, as such it is the natural progression to perpetuate scenario-based V\&V testing for AVs. As such, this dissertation proposes an end-to-end scenario-based V\&V testing framework for AV V\&V testing, which intelligently and effectively identifies and proposes solutions to the challenges of testing AVs and other complex systems under time and resource constraints. The dissertation begins by identifying sources of AV V\&V scenarios, then constructing modular and measurable scenarios that are validated through scenario-based tests. Then, challenges in scenario-based testing of highly complex, high-dimensional scenarios are identified through modeling. This dissertation includes numerous studies of sequence sampling and exploration strategies to efficiently explore high-dimensional spaces, and to select high-value and edge scenarios. After the foundations of the proposed framework are validated, the dissertation diversifies: a first-of-it's kind AV scenario-description language for network-based traffic simulation is proposed, a practical means of estimating scenario complexity of AV V\&V scenarios is proposed, and interpretation of the scenario test data by leveraging explainable artificial intelligence is proposed. Finally, the framework is stress tested via highly complex, high-dimensional scenario tests to evaluate the scalability, interpretability, and performance of the dissertation research.