KIZAM
Guideline for the funding of research and development projects under the BMWK New Vehicle and System Technologies programme
Project name:Artificial Intelligence in Goals and Requirements Management
Project duration: 01.07.2021 – 30.06.2024
Funding ID: 19I21029C
Location: Munich and Aachen, Germany
Project sponsor: TÜV Rheinland Group
Project leaders: BMW Group
Grantor: Bundesministerium für Wirtschaft und Klimaschutz – BMWK
RESEARCH OBJETIVES
OBJETIVE 1
Identification of AI potential in requirements management (RM)
OBJETIVE 2
Intelligent requirements testing: Correlations and contradictions between requirements are identified by machines
OBJETIVE 3
Increasing efficiency throughout the entire product development process
DESCRIPTION
Identifying the Project’s Challenge
The requirements for the vehicle development process are highly complex. The definition of requirements, such as specifications, serves as the starting point for all subsequent functional, operational, and solution principles. As a result, quality is of enormous importance in terms of unambiguous, complete, and non-redundant requirements. Thus far, traditional databases have been used to compensate for the increasing requirements due to the number of variations and complexity, but these methods are becoming increasingly limited. To support future digital development processes, all complex interactions and requirements must be described without the need for manual and time-consuming checks.
The KIZAM project aims to research the use of AI methods in requirements management for vehicle development. This involves developing machine-readable, optimized, and consistent requirements in vehicle development through adapted software solutions and AI algorithms. These algorithms build on existing systems and solutions and can be easily and quickly adapted if the project is successful. Moreover, the use of AI in requirements management allows for the fast and efficient comparison of all relevant requirements data, identifying interactions even across requirements levels, process interfaces, and system boundaries. This significantly improves the quality of the requirements catalogue.
PEM Motion’s Approach and What Has Been Achieved So Far?
PEM Motion is responsible for three different use cases focused on homologation-specific requirements and audit-proof requirement confirmation in human-machine interaction. A procedure is developed for one use case and then transferred to the other use cases during the project. PEM Motion is responsible for the following tasks:
- Developing use cases
- Analyzing the automation potential of the requirements
- Investigating the potential for supporting personnel-intensive processes for homologation and safety certification through automated requirements testing
- Parameterizing requirements
- Providing exemplary requirements from the use cases
- Deriving standardized requirements
- Developing a standardized classification rule for requirements
- Automating the structuring of requirements
- Evaluating the applicability of the methodology for the use cases.
In the course of the project so far, the mapping and analysis of the requirements management (RM) processes for all use cases have been completed. In addition, process steps have been identified where AI can potentially provide support, and a process model has been developed to uniformly describe the RM processes. Suitable AI methods have been developed and tested to provide targeted support for the RM processes. Moreover, different language models for the AI methods were analyzed.