Reducing cycle time and costs of embedded control software using rapid prototyping and automated code generation and test tools

摘要

Electronics content in off-highway machinery and vehicles is growing in size and complexity. Mr. Ronald K. Leonard, 1998 SAE President, recently gave a lecture titled "Electronics - A New Engine For Agricultural Machines" in London, UK that described electronics as an enabler of new functionality and business solutions. Electronics is the term for both hardware and embedded software that enables manufacturers to optimize and expand performance of traditional mechanical and hydraulic machinery. This paper focuses on the development of the software that is involved in controlling devices such as an engine, a transmission or a hydraulic backhoe. The off-highway market is now looking at ways to improve productivity using advanced tools and methods. The tools available to the off-highway industry were first used by aerospace companies ten years ago. The automotive industry has just started to adopt these methods and the off-highway industry seems to be close behind. Previous software engineering methods in these industries required hand generation of embedded software from high-level and often incomplete requirements supplied by system control engineers. This resulted in software life-cycle inefficiencies that delayed releases and caused maintenance difficulties. This paper presents a model-based approach to developing Electronic Control Units (ECUs) from requirements capture through validation of the production ECU. With this approach, the constraints of the actual target environment are confined to the early development stages of simulation and rapid-prototyping, resulting in software requirements that are then automatically converted into production-ready code with little iteration and minimal effort. In summary, reducing cycle time and costs can be achieved in the development of software using a model-based development environment by automating process steps (including activities never automated before). This allows the removal of defects early in the design and test process as opposed to removing them after the costly integration and system testing has begun, which is much more costly.