405x Filetype PDF File size 1.59 MB Source: www.tuvsud.com
Virtual homologation
of an ALKS according
to UNECE R157
|
Co-authored by
White paper
Abstract
Proving that an automated driving risk is acceptable requires extensive investigation into the challenge
of handling real-world traffic. UNECE R157 is the first regulation to formalise the idea of virtualising major
parts of the validation and homologation process. This white paper discusses a recommended process
for the virtual homologation of an Automated Lane Keeping System (ALKS) by following the provisions in
UNECE R157, including test approaches to prove system safety.
TÜV SÜD
Contents
INTRODUCTION 3
PROVING SYSTEM SAFETY 6
RELIABILITY OF SIMULATION RESULTS 10
CONCLUSION 14
About TÜV SÜD expert
Christoph Miethaner
Test Methods Engineer for Highly Automated Driving at TÜV SÜD
Christoph is an automated driving specialist with four years of industry experience and
participates in various committees that promote the advancement of these complex
automotive technologies. He is a member of the openSCENARIO working committee which
focuses on the virtual development, test and validation of driver assistance functions,
automated and autonomous driving. He is also a member of the committee ISO/TC22/SC33/
WG09 for test scenarios of automated driving systems, and DIN NA 052-00-33-09 AK,
which is the national mirror committee of ISO/TC22/SC33/WG09. He is a member of the
Methods, Models and Effectiveness Calculation working committee at P.E.A.R.S (Prospective
Effectiveness Assessment for Road Safety), and represents the international not-for-profit
association CITA (international motor vehicle inspection committee) at UNECE VMAD
(Informal Working Group on Validation Methods for Automated Driving at United Nations
Economic Commission for Europe).
About dSPACE expert
Jann-Eve Stavesand
Head of dSPACE Consulting
Jann-Eve supports customers worldwide to define test strategies for complex automotive
systems, helping them to overcome the challenges of developing safety-critical systems.
He is a specialist in testing and quality assurance of complex automotive software and
systems, including their approval and homologation. He sits on various standards committees
in Germany and was involved in the development of ISO 26262:2018 - "Road vehicles –
Functional safety", focusing on the software and processes elements. He is also involved
with the standardisation of Safety of the Intended Functionality (SOTIF).
2 Virtual homologation of an ALKS according to UNECE R157 | TÜV SÜD
Introduction
The high complexity of an automated Proving that a risk is acceptable According to these guidelines,
driving task has prompted the idea requires extensive investigation into simulation plays a crucial role for
of virtualising major parts of the the challenge of handling real-world safety argumentation. Validating
validation and homologation traffic. Currently, the prevailing certain safety measures by simulation
process. UNECE regulation R157, opinion among experts is that this task means that it must also be proven
for the approval of vehicles with can only be performed on a virtual why the results of the simulation are
ALKS, is the first regulation to basis. Therefore, several types of reliable. The section “Reliability of
formalise these ideas. The authors testing are required (from simulation simulation results” gives an overview
of UNECE R157 not only defined to real-world tests, see figure below). of the different methods to achieve
a fixed set of test cases, but also The following sections describe trustworthiness of a virtual test setup.
augmented the prescribed set of the virtual evaluation method used In addition, we compared simulated
tests by making it a requirement to to determine whether a risk is traffic situations with equivalent
achieve higher-level principles, such acceptable for an ALKS, as per the real-world test drives to analyse the
as acceptable risk. UNECE R157 guidelines. accuracy of the simulation.
CERTIFICATION AND SAFETY ARGUMENT PROCESS
AND BASIS FOR VERIFICATION
Argument for certification authorities
Regulations Argument for safety
Methods Verification results Exploration of unknown
Field observationT
Selection Selection rustworthiness
KPI
Scenario
catalogue
Verification Validation
Scenario approach
Test management framework (end-to-end)
ODD
Design Test execution plan /
V&V Strategy
Risk Analysis
TÜV SÜD | Virtual homologation of an ALKS according to UNECE R157 3
Status quo – ALKS UNECE R157
A key aspect of UNECE R157 that which defines the same term as Compared to the traditional
is unique to the regulatory and “risk judged to be unacceptable in homologation process, where
standardisation framework is the a certain context according to valid real-world test drives at the end
definition of a reasonable risk. societal moral concepts,” UNECE of the development process form
UNECE R157 Annex 4 Clause 2.16. R157 redefines the threshold for an the basis for homologation, UNECE
defines an unreasonable risk as acceptable risk to a comparison that R157 demands independent audits
“[...] the overall level of risk for can be measured and argued to a of relevant processes and their
the driver, vehicle occupants and certain degree. This is a key enabler implementation. Furthermore,
other road users which is increased for the approval of vehicles with several assessments are required
compared to a competently and higher levels of automation, since throughout the product lifecycle. For
carefully driven manual vehicle.” it acknowledges the presences of a details, refer to the figure below [1].
Compared to existing sources, such residual risk and provides guidance
as ISO 26262-1:2018 Clause 3.176, on how this risk can be quantified.
CONCEPT FOR CERTIFICATION – THE THREE PILLARS:
PHASES OF TYPE APPROVAL
MAIN POINTS FOR CERTIFICATION
Real-world ■ Overall impression of system
test drive behaviour on public roads
■ Assessment of system's ability to
cope with real-world traffic
■ Validation of audit/assessment
Physical results with real-world behaviour
certification ■ Assessment of system behaviour in
tests fixed set of challenging cases
■ It is possible to reproduce situations
■ Audit of development process and
Audit V&V strategies
■ Assessment of safety concept
(functional safety, safety of use)
Assessment and measures taken
■ Use of simulation results
Simulation ■ Measuring process maturity
4 Virtual homologation of an ALKS according to UNECE R157 | TÜV SÜD
no reviews yet
Please Login to review.