Options
2018
Master Thesis
Title
Dynamic risk assessment at intersections
Abstract
Autonomous driving has a great potential for reducing traffic accidents. This paves a way for the improvement of road safety. However, the very notion of risk is not always a clearly defined concept. Existing work have shown how risk metrics used for performing the risk assessments are domain specific and are defined on a vehicular level. On a global level for an intersection with multiple vehicles, these risk metrics cannot be applied without further modifications. First part of the work involves creating a test scenario using SUMO simulator, importing this scenario into the Risk Metric Calculator (RMC), deciding on the prediction models that will be used for carrying out the risk assessment and defining a new mission level risk metric to carry out a Dynamic Risk Assessment at the considered intersection. To this end, intended trajectory prediction model and reachable area prediction model are used for risk assessment and a new risk metric MTTC (Mission Level Time to Collision) is defined. The concept of Conditional Safety Certificates (ConSerts) is applied in the second half of the work. The result of evaluation of these ConSerts at runtime is used to determine if the Dynamic Risk Assessment (DRA) on different confidence levels at the intersection could be performed or not. To achieve this a communication between the SUMO simulator and ConSerts evaluation and visualization services has been created.
Thesis Note
Kaiserslautern, TU, Master Thesis, 2018
Publishing Place
Kaiserslautern