Integration of Automatic Code Generation in Model-Based Algorithm Development; Prerequisites, Workflow and the Human Factor

摘要

Automatic code generation is an established technology in automotive and aerospace industries that is also adopted in commercial vehicle embedded software development. Code generation brings the productivity gains of model-based algorithm design to the next stage of the process - production software development. The technology has matured to the point where it satisfies most technical and usability requirements and the success of code generation depends on judicious deployment, efficient work practices and acceptance by all users involved. It has been demonstrated that model-based software development and code generation in particular can shorten development cycles while staying true to the requirements and maintaining software quality through multiple algorithm iterations. These gains, though, only come as a result of careful combination of tools and methods. This paper discusses solutions to common organizational and technical challenges of model-based software development. These include workflows to bridge disconnects between algorithm and software development, supplier and OEM, companies and their foreign subsidiaries. In addition to organizational planning, technical infrastructure is needed to provide proper data management, change tracking, merging of parallel versions, integration with legacy software, reuse of test vectors, linking to requirements and enforcing modeling style guidelines. Examples of successful workflows, tool choices and best practices are described. The importance of the human factor is also acknowledged.