Developing Vehicle Configuration Rules

Vehicles are sold in many variants with different engine horsepower, wheel dimensions, type of headlights etc. Vehicle customers specify individual vehicles by the selection of such features during the so-called “sales configuration” process. Logic expressions called vehicle configuration rules are often utilized for automating the sales-to-delivery process. The development of vehicle configuration rules includes authoring and evaluation of the configuration rules. The goal is the make sure that the configuration rules specify so-called “valid” configurations. Valid in this context is a user-defined state based on perception of domain specialists, e.g. design engineers of brakes. The aim of this thesis is to develop methods for configuration rules development that efficiently ensure that configurations are buildable by making the configuration rules development process more time-efficient and less error-prone. The problem is that the industrial visualization tools for developing configuration rules, henceforth CR visualization tools, are argued to be difficult to use. Consequently, the users find it difficult to validate configuration rules. The problem is in this thesis approached from scratch by first suggesting a generalized information model for vehicle configuration rules. The information model was derived from industrial studies and literature reviews. Then, user studies at three vehicle manufacturing companies were conducted in order to formalize the authoring methods and for study existing CR visualization tools. Limitations of current CR visualization tools were identified, which were addressed in a new CR visualization tool. This new tool uses one single user interface, which eliminates the swapping between windows. Also, the alternative authoring methods and potentially missing items are visualized. Some activities when evaluating the configuration rules are thereby facilitated. The new CR visualization tool has been iteratively developed and evaluated through formative usability tests. The test results were positive in terms of real users correctly conducting test tasks, appreciating the new CR visualization method as well predicting a more time-efficient CR development process. The planned future work includes more usability tests to study if there are any unforeseen usability threats