Search Results

Contemporary cutting-edge technologies, such as automated driving brought up vital questions about safety and relativized the safety assurance and acceptance criterion on different aspects. New risk assessment, evaluation, and acceptance justifications are required to assure that the assumptions and benchmarking are made on a reasonable basis. While there are some existing risk evaluation methods, most of them are qualitative in nature and are subjective. Moreover, information such as the safety performance indicators (SPIs) of the sensors, algorithms, and actuators are often not utilized well in these methods. To overcome these limitations, in this paper we propose a risk quantification methodology that uses Bayesian Networks to assess if the residual risk is reasonable under a given scenario.

Scenario-based testing is one of the widely used approaches to verify the intended functionality of autonomous vehicles. In this approach, the vehicle is tested by exposing it to various known scenarios and then analyzed based on corresponding results. While scenarios can be constructed in simulation, they are hard to control in the real-world. Hence, a coverage metric such as scenario coverage is used to understand how many of the known scenarios has the vehicle been exposed to in simulation or real world. However, scenarios can be detailed at various levels of abstraction and many situations are possible to occur within a scenario. Therefore, considering only scenario coverage may not ensure that all the dependencies among components in the system and the ODD factors within a scenario are taken into account and verified sufficiently.

Two major steps involved in SOTIF analysis are defining acceptance criteria and estimating the validation target. While acceptance criteria aids in determining if we have an acceptable residual risk corresponding to a hazardous scenario, the validation target specifies the amount of testing effort (in hours or representative miles) that is needed to ensure that the acceptance criteria are met. The current approaches for defining acceptance criteria heavily rely on existing fatality databases or naturalistic driving study data sets. The criterion is selected based on average number of fatalities or crashes per mile or per one hour of operation. The validation target is then calculated based on acceptance criteria. However respective validation targets., are these values really reflecting the acceptable risk criteria and targets?