Automated driving systems(ADS) are major trend and the safety of such critical system has become one of the most important research topics. We usually use scenarios in order to define the specifications of ADS. In these scenarios, graphical diagrams are often used to represent abstractly the positioning and behavior of vehicles. However, such diagrams are not suitable for the development of high-reliability systems, because they are informal and may cause discrepancies among different engineers. In this paper, we propose a formal speci-fication language called Bounding Box Specification Language (BBSL) which allows us to write rigorous specifications of ADS. BBSL describe multiple types of objects in a driving environment, such as vehicles and pedestrians, as bounding boxes defined as two-dimensional interval, and describe positional relationships between them in mathematical notation. It is capable of strictly delineating many positional relationships while being also capable of expressing specifications that are concise enough to be read and written manually. Therefore, BBSL is suitable for describing the specification of Object and Event Detection and Response (OEDR) among the tasks of ADS. In this paper, we describe what kind of description BBSL enables, and describe its operations. Then, we show examples of specifications of ADS written in BBSL and discuss the advantages of specifications written in BBSL.

Recently, machine learning, particularly deep learning, is attracting much interest and is applied in various systems. Applications include not only entertainment systems, but safety-critical systems such as those found in autonomous vehicles. The reliability of such safety-critical systems must be guaranteed before they are released into society. However, methods for ensuring the safety of machine learning-based systems have yet to be established. In this paper, we propose a method for systematically evaluating the safety of such systems. The method consists of dataset-based safety analysis and statistical evaluation of testing results. In the safety analysis, we extend the widely used fault tree analysis to deal with datasets. In the testing, we use statistical estimation to guarantee recognition rates obtained in the safety analysis. We conducted experiments using a handwritten character recognition system implemented as a CNN to demonstrate the feasibility and effectiveness of our method.