Plenary Lecture

Software Tests for Model Based Applications in the Automotive Industry

Professor Andrzej W. Ordys
Co-author: Peter Kock
Faculty of Engineering
Kingston University in London
Friars Avenue, Roehampton Vale
London SW15 3DW, UK
a.ordys@king.ac.uk

Abstract: In automotive industry, computer control, fault detection and communication are becoming increasingly complex and interrelated with each other. This prompts development of formal methods for software design and verification.
The model based development of automotive functionality has become extremely popular especially in drive train and engine applications. The approach appears to make the development more modular, with the programming code being more reusable and easier to test. This article looks at the advantages and disadvantages of this approach. It concerns the range of tests, from first component tests through integration and system tests to the final acceptance tests.
It discusses model based development using the V model and shows some differences to code based development.
The first component tests are generally divided into static tests (Reviews, Rule checkers, Metrics, Static analysis) and dynamic tests (model in the loop, software in the loop, processor in the loop). In addition, an important aspect is verification of models using non-floating point arithmetic, as this method of calculation of often used in automotive embedded computers (e.g. FPGA). The formal verification of such models is discussed. With respect to the integration tests, the article concentrates on some issues of hardware-in-the-loop testing. The system tests are also discussed, especially the structure of system information in a model based approach and the preconditions determined by hardware in the loop tests. In the final acceptance tests, the main problems contain: formal link between the system requirements and the test methods and coordination and optimisation of the whole test process.
Current trends in automotive industry are toward higher level of integration of the development, verification and manufacturing of systems. Therefore, the traceability is also discussed, with a mention of a model of an integrated tool chain. The article shows the pit falls of tool orientation and a way out of the disaster.
An example of software development and testing process in a leading automotive company will illustrate the considerations.
Finally, the article looks at the future of software development in automotive industry, will development of formal methods and increasing standardisation lead to a closer cooperation of different manufacturers?

Brief Biography of the Speaker:
A. W. Ordys is Professor of Automotive Engineering at Kingston University, London. Previously he worked at the University of Strathclyde as the British Energy Senior Lecturer in Control Systems in the Department of Electronic and Electrical Engineering.
His current research interests include control and optimisation of engine power train, Design and implementation of embedded systems in automotive applications, hybrid control of fast and high-precision mechanical systems and performance assessment of control systems.
He has been contributing to the theoretical development of stochastic and predictive control theory and the theory of benchmarking control algorithms.
He has contributed to a range of industrial application projects including those for the automotive, power, oil and gas, chemical, defence and metal processing industries. Examples of industrial collaborators include: General Motors, MAN, British Energy, National Instruments.
Dr. Ordys is Member of IEE and Chartered Engineer, and Senior Member of IEEE. He is a past Chairman of EC&I Section of IEE Scotland.
He is a member of the Executive Committee of IET Control and Automation Professional Network.