SMILE III

Period

1 April 2020–31 March 2022

Project manager

Cristofer Englund

Safety analysis and verification/validation of MachIne LEarning based systems (SMILE III)

The purpose of SMILE III is to further develop and demonstrate the safety-cage concept developed in previous projects. The goal is to be able to safely use ML-based functions in a safe way in safety-critical vehicle systems. The goal is also to use the safety-cage concept to understand how future safety standards affect the development and management of ML-based systems. The project will also develop an architecture for how safety cage and perception modules can interact.

The project is based on collaboration between industry, institutes and academia. The participants work closely with the project's tasks and deliverables. The ambition is that the results can be implemented in both simulation models and in miniature vehicles and show how the developed concept can ensure safe use of ML-based functions in vehicles.

The project is divided into 6 work packages. Project management, Architecture, Safety Case Strategy, Safety-cage design and optimization, Data management and V&V strategy and Autonomous platform. The project is led by RISE and each work package has a leader who drives and coordinates the work within resp. wp.

Short facts

Research area

Systems for accident prevention and AD

Financier(s)

VINNOVA/ FFI

Partners

RISE

Combitech

Semcon

ESI Nordics

INFOTIV

Project type

SAFER connected project

Related publications

An autonomous performance testing framework using self-adaptive fuzzy reinforcement learning

Test automation brings the potential to reduce costs and human effort, but several aspects of software testing remain challenging to automate. One such example is automated performance testing to find performance breaking points. Current approaches to

Authors : M. H. Moghadam, M. Saadatmand, M. Borg, M. Bohlin, and B. Lisper

Year : 2022

Published in : Software Quality Journal

Machine Learning Testing in an ADAS Case Study Using Simulation-Integrated Bio-Inspired Search-Based Testing

This paper presents an extended version of Deeper, a search-based simulation-integrated test solution that generates failure-revealing test scenarios for testing a deep neural network-based lane-keeping system. In the newly proposed version, we utilize a

Authors : Mahshid Helali Moghadam, Markus Borg, Mehrdad Saadatmand, Seyed Jalaleddin Mousavirad, Markus Bohlin, Björn Lisper

Year : 2022

Published in : Journal Of Software: Evolution And Process

Exploring ML testing in practice -- Lessons learned from an interactive rapid review with Axis Communications

There is a growing interest in industry and academia in machine learning (ML) testing. We believe that industry and academia need to learn together to produce rigorous and relevant knowledge. In this study, we initiate a collaboration between stakeholders

Authors : Qunying Song, Markus Borg, Emelie Engström, Håkan Ardö, Sergio Rico

Year : 2022

Published in : Proc. of CAIN 2022 - 1st International Conference on AI Engineering - Software Engineering for AI

Ergo, SMIRK is Safe: A Safety Case for a Machine Learning Component in a Pedestrian Automatic Emergency Brake System

Integration of Machine Learning (ML) components in critical applications introduces novel challenges for software certification and verification. New safety standards and technical guidelines are under development to support the safety of ML-based systems

Authors : Markus Borg, Jens Henriksson, Kasper Socha, Olof Lennartsson, Elias Sonnsjö Lönegren, Thanh Bui, Piotr Tomaszewski, Sankar Raman Sathyamoorthy, Sebastian Brink, Mahshid Helali Moghadam

Year : 2022

Published in : Software Quality Journal

Public Report SMILE III

Future autonomous vehicles will rely on Deep machine learning algorithms (DML), whose correct behavior cannot be guaranteed by traditional software engineering approaches. This uncertainty is a big obstacle to ensure the quality of systems using DML

Authors : Markus Borg, Thanh Bui, Jens Henriksson, Martin Karsberg, Olof Lennartsson, Gustaf Bergström, Sebastian Brink, Sankar Sathyamoorthy, Erik Abenius

Year : 2022

On Improving Validity of Deep Neural Networks in Safety Critical Applications

Context: Deep learning has proven to be a valuable component in object detection and classification, as the technique has shown an increased performance throughput compared to traditional software algorithms. Deep learning refers to the process, in which

Authors : Jens Henriksson

Year : 2020

Published in : Chalmers University of Technology