Organization of machine learning based product development as per ISO 26262 and ISO/PAS 21448

Radlak, Krystian; Szczepankiewicz, Michał; Jones, Tim; Serwa, Piotr

doi:10.1109/prdc50213.2020.00022

Cited by 11 publications

(7 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…From the left side, each column is Eqs. ( 9), ( 10), (11), and (12), respectively. In this result, the corruption number is 0 for the Gaussian noise and 5 for the severity.…”

Section: Fldmmentioning

confidence: 99%

Robust Object Detection Against Multi-Type Corruption Without Catastrophic Forgetting During Adversarial Training Under Harsh Autonomous-Driving Environments

Kim

Shin

2023

IEEE Access

View full text Add to dashboard Cite

It is important to build robust object detector (ROD) in real-world applications because snow, rain, fog, motion blur, and various kinds of corruption can occur in autonomous-driving environments. Adversarial training (AT) is one of the best solutions to build a robust deep neural network. However, applying AT has a risk of sacrificing clean performance, even though robustness is improved, which is called catastrophic forgetting (CF). In particular, CF in an autonomous-driving environment is more challenging for two reasons. The first is CF is worsened due to various types of corruption. The second is the degradation of clean performance can lead to a risk of overall performance degradation because more than 60% of the total data is clean (based on Bdd100k). Therefore, we propose an ROD framework to ensure not only robustness against corruption but also prevent degradation of clean performance, despite the two aforementioned difficulties. The ROD framework utilizes a training methodology with an adversarial defense module (ADM) based on the intermediate representative feature (IRF) concept. This framework can improve robustness without CF under multi-corruption environments. In this paper, we report on three main achievements. The first is that the mean performance under corruption (mPC) of RetinaNet was improved by 32.14% with an mAP degradation of only 0.2%, based on COCO 2017. The second is that our method achieved state-of-the art results with 86.8% relative performance under corruption (rPC) compared to Hybrid Task Cascades with 64.6% rPC on the ROD benchmark. The third is that our ROD methodology achieved 32.29% and 31.54% mPC at 15-type seen corruption and four-type unseen corruption, respectively. The ROD framework is also applied to an autonomous-driving domain showing that it operates well, even under harsh environments in the Bdd100k dataset.

show abstract

“…From the left side, each column is Eqs. ( 9), ( 10), (11), and (12), respectively. In this result, the corruption number is 0 for the Gaussian noise and 5 for the severity.…”

Section: Fldmmentioning

confidence: 99%

Robust Object Detection Against Multi-Type Corruption Without Catastrophic Forgetting During Adversarial Training Under Harsh Autonomous-Driving Environments

Kim

Shin

2023

IEEE Access

View full text Add to dashboard Cite

show abstract

“…The safety engineering phase of autonomous vehicles was applied to ISO26262 and ISO/PAS 21448, but the process was designed from a safety perspective rather than the system development process and it has limitations in application [27]. Although ML based study was proposed according to ISO 26262 and ISO/PAS 21448, there is a limit to the lack of research and analysis on ISO/PAS 21448 [28]. Precedent studies have shown that there are many studies related to ISO26262 and ISO/PAS 21448, but there are not enough studies to integrate the two processes.…”

Section: The Need For Integration Of Iso 26262 and Iso/pas 21448mentioning

confidence: 99%

Development of ISO 26262 and ISO 21448 Concept Design Process Integration

Yang¹,

Li²,

Liu³

et al. 2022

Proceedings of the 2022 International Conference on Computer Science, Information Engineering and Digital Economy (CSIEDE 2022)

View full text Add to dashboard Cite

The development of technology has led to the improvement of vehicle safety and the introduction of autonomous driving system. The autonomous driving related system requires various parts such as camera sensors. Various parts are being added to the vehicle, which is making the system more complicated. Safety-related issues are becoming more important to prevent accidents that may occur during autonomous-driving due to errors caused by increased complexity of the system. For the safety of autonomous driving systems, the International Organization for Standardization introduced the ISO 26262 standard to prevent accidents caused by errors in electrical/electronic systems applied to vehicles. Recently, ISO 21448 is being introduced to compensate for the areas that ISO 26262 does not deal with in relation to autonomous driving. In order to implement autonomous driving technology, ISO 21448 is needed because it is necessary to prevent the risk of performance constraints while implementing functions. However, ISO 21448 is a standard under enactment, and it is difficult to apply to the actual development process. Also, ISO/PAS 21448 is applied with ISO 26262 in the actual development process, so consideration for the application of both processes is needed. This paper proposes an integrated process for applying ISO/PAS 21448 to the actual development process. The integrated process is a combination of ISO 26262 and ISO/PAS 21448, and the concept part of the process is strengthened. For the integration of ISO 26262 and ISO/PAS 21448, each process was analyzed and newly constructed to enhance applicability and reduce trial and error of concept development process.

show abstract

“…In [12] both ISO standards are combined to create a product development process for ML. The authors incorporate ISO/PAS 21448 into ISO 26262 work products and development phases.…”

Section: Related Workmentioning

confidence: 99%

Safety Assurance of Machine Learning for Chassis Control Functions

Burton

Kurzidem

Schwaiger

et al. 2021

Lecture Notes in Computer Science

View full text Add to dashboard Cite

This paper describes the application of machine learning techniques and an associated assurance case for a safety-relevant chassis control system. The method applied during the assurance process is described including the sources of evidence and deviations from previous ISO 26262 based approaches. The paper highlights how the choice of machine learning approach supports the assurance case, especially regarding the inherent explainability of the algorithm and its robustness to minor input changes. In addition, the challenges that arise if applying more complex machine learning technique, for example in the domain of automated driving, are also discussed. The main contribution of the paper is the demonstration of an assurance approach for machine learning for a comparatively simple function. This allowed the authors to develop a convincing assurance case, whilst identifying pragmatic considerations in the application of machine learning for safety-relevant functions.

show abstract

Organization of machine learning based product development as per ISO 26262 and ISO/PAS 21448

Cited by 11 publications

References 22 publications

Robust Object Detection Against Multi-Type Corruption Without Catastrophic Forgetting During Adversarial Training Under Harsh Autonomous-Driving Environments

Robust Object Detection Against Multi-Type Corruption Without Catastrophic Forgetting During Adversarial Training Under Harsh Autonomous-Driving Environments

Development of ISO 26262 and ISO 21448 Concept Design Process Integration

Safety Assurance of Machine Learning for Chassis Control Functions

Contact Info

Product

Resources

About