A scenario database might be available for test and validation purpose. The scenario data might fit the requirements and the ODD. Then it needs to be checked if the resulting scenario quality is sufficient for testing purpose and if test cases can be derived from the scenarios.
In the design process of ADFs different thresholds have to be determined in order to provide an acceptable ADF behaviour. Usually this behaviour is closely related to human behaviour in non-critical driving. In order to define the thresholds and parameters, data from scenario databases can be applied in case it represents naturalistic driving (Weber, H. et al., 2023).
Main Question
Is the data quality of scenarios sufficient?
Sub-Questions
- Are requirements on data quality known?
- Are the specifications on data quality known?
- Is naturalistic driving behaviour available in the scenarios?
- Are enough scenarios available to provide statistical representativeness?
References
- Weber, H. et al. (2023) ‘Holistic driving scenario concept for urban traffic’, 2023 IEEE Intelligent Vehicles Symposium (IV). doi: https://doi.org/10.1109/iv55152.2023.10186385
- Glasmacher, C. et al. (2023) ‘Acquire driving scenarios efficiently: A framework for prospective assessment of cost-optimal scenario acquisition’, arXiv [cs.SE]. 26th IEEE International Conference on Intelligent Transportation Systems ITSC 2023, Bilbao, Spain, 24. – 28.09.2023. Available at: http://arxiv.org/abs/2307.11647 (Accessed: 05 February 2024)
- Scholtes, M. et al. (2021) ‘6-layer model for a structured description and categorization of urban traffic and environment’, IEEE Access, 9, pp. 59131–59147. https://doi.org/10.1109/access.2021.3072739