ページタイトル

Project InfoProject Info

ページタイトル画像
エディタV2

About SAKURA

SAKURA

  • Socially acceptable and technically sound safety assurance methodologies are needed to safely introduce Automated Driving systems into the market.
  • The SAKURA project (Safety Assurance KUdos for Reliable Autonomous vehicles) is a large scale coordinated initiative funded by the Japanese Ministry of Economy, Trade and Industry (METI) that aims at harmonizing data acquisition, developing research methodologies and coordinating standardization activities through joint efforts by vehicle manufacturers and traffic safety research institutions.

SAKURA RESEARCH PROJECT

ENGINEERING FRAMEWORK FOR AD VEHICLE TEST SCENARIOS

A schematic of the overall safety assurance developed process is shown in Figure 1. The schematic is based on the project management V-model typically applied to develop connected vehicles, advanced driver assistance systems (ADAS) and AD systems [1]. The process covers all the product development stages from planning, design, implementation, evaluation (verification and validation), to release.

Figure 1
Figure 1 Overall scheme of the safety assurance process

[1] Themann P, Raudszus D, Zlocki A, Eckstein L. Holistic Assessment of Connected Mobility and Automated Driving. ATZ worldwide. 2016;118(1):26-31.

SOCIAL CONTEXUALIZATION OF THE ENGINEERING FRAMEWORK

The Japanese government recently released a technical safety guideline for AD systems, which approximately read ”Within the Operational Design Domain, AD vehicles shall not cause injurious or fatal accidents that are reasonably foreseeable and preventable” [2] [3] [4]. By contextualizing the AD systems safety assurance engineering framework proposed with respect to the governmental safety guideline, it is possible to propose a methodology that facilitates social acceptance of the AD systems. A scheme of how such contextualization may be achieved is shown in Figure 2. The proposed scheme follows a top-down approach in which the ODD (Operational Design Domain) is defined considering the top safety goal, the test scenarios are developed based on real-world traffic data and the virtual and physical testing is conducted under the same framework. Therefore, the safety goals and engineering framework can be harmonized between all social and industrial stakeholders, while the development of the tools and methodologies can be led by the industrial stakeholders.

Figure 2
Figure 2 Top-down approach for social contextualization of the engineering framework for AD safety assurance

[2] Ministry of Land, Infrastructure, Transportation and Tourism (2018). Automated Vehicles technical safety guidelines. September; 2018.
[3] European Commission (2018). Guidelines on the exemption procedure for the EU approval of automated vehicles.
[4] UNECE World Forum for Harmonization of Vehicle Regulations (WP.29) (2019). Framework document on automated/autonomous vehicles.

TEST SCERANIO GENERATION PROCESS FOR AD SAFETY ASSURANCE

The approach to test scenarios for AD vehicle safety assurance is shown in the Figure 3. First, scenarios are structured according the three primary driving functions: recognition, judgment and operation in order to cover holistic root cause from the viewpoint of the physics of AD systems. Second, for each structured scenario, parameter items and their ranges for foreseeable conditions are defined based on real-world traffic monitoring data. Third, among the foreseeable conditions, the ranges are narrowed down to those that correspond with preventable conditions. Both the evaluation of the completeness of the structured scenarios as well as the steps to define ranges for foreseeable and preventable conditions are developed based on real world data.

Figure 3
Figure 3 Test Scenario Generation Process for AD Safety Assurance

Logical scenario[5]: A model of the time sequence of scenes whose parameters are defined as ranges, which begins with an initial scene and, at a defined point in time
Concrete scenario[5]: A parameterized model of the time sequence of scenes (logical scenario), which begins with an initial scene and at a defined point in time

[5]: DIN/SAE(2018), Terms and Definitions Related to Testing of Automated Vehicle Technologies.

MAIN GOALS OF SAKURA PROJECT

  • To develop an automated vehicle system safety assurance engineering process that accounts for all foreseeable safety relevant scenarios, with a particular focus on motorways.
  • To research and develop the fundamental technology necessary to enable the developed safety assurance process.
  • To lead international standardization activities towards the harmonization of scenario structure, parameter range, and safety criteria establishment towards globally accepted common approaches.