Knowledge Management in Research and Development Projects: Mapping intellectual Assets for Innovation in Manufacturing

Abstract

Innovation is the key driver of competitive advantage in a knowledge economy. In the manufacturing industry, a major portion of innovation emerges from research and development (R&D) projects. These projects require substantial resources, yet their outcomes remain uncertain. While previous research has explored ways to enhance R&D outcomes, certain aspects remain unexamined.

Intellectual assets—data, information, knowledge, and wisdom (DIKW)—play a role in these projects, but their specific involvement is not well understood. Despite the abundance of data and information, much of it is either underutilised or misapplied due to a lack of clarity regarding its utility. Furthermore, the role of different knowledge types in innovation projects remains unclear, as knowledge is often treated as an abstract monolithic asset. This thesis examines knowledge management in R&D projects within the manufacturing sector.

The studies in this thesis were mainly conducted by the case study methodology with participant observation. Two R&D projects are analysed: the first focuses on developing an ergonomic assessment system using an inertial motion capture suit, while the second investigates indoor positioning of metal parts using a passive RFID system. Each project is divided into chronological stages, with each stage representing a major decision or action. Additionally, surveys and user studies were conducted in the studies.

The first study proposes a set of criteria to distinguish and classify DIKW, identifying instances of each as inputs and outputs at various project stages. These criteria are validated through an online survey. The mapping of DIKW reveals that information and knowledge are present in nearly all stages, whereas data and wisdom appear in only a few. Wisdom is essential for initiating an R&D project and making critical decisions, such as modifying objectives or closing the project. Data is primarily generated and utilised during development, testing, and equipment trials. The data generated in R&D can also provide insights into data production in manufacturing operations, aiding in the selection of valuable data and the allocation of computing resources for analysis. Since wisdom and knowledge influence the quality of data and information, a more detailed investigation of knowledge types was necessary.

The second study maps and categorizes knowledge into its types based on selected typologies from the literature. A review of various knowledge typologies identifies key dimensions of knowledge. A survey with scholars suggests that Zack’s (1999) five functional knowledge types provide the most effective framework for knowledge mapping of R&D projects in manufacturing, while Polanyi’s (1958) binary typology remains relevant due to its simplicity. Knowledge management tools structured around functional knowledge types can improve the capture, organisation, and retrieval of knowledge for reuse in other projects.

The third study evaluates the usability of Obsidian, a personal knowledge management application, for knowledge sharing in R&D projects. Among the tested modes of access, the search modes are the most efficient for retrieving specific knowledge, while participants prefer the FAQ graph view and report outline for exploring project knowledge bases, as these provide an overview that facilitates navigation. Tools like Obsidian have significant potential for R&D knowledge management, provided they are enhanced with additional features and appropriate training. Specifically, plugins that structure knowledge according to type could be developed, and users could optimise knowledge retrieval by combining different modes of access.

Although the proposed DIKW criteria could be refined, researchers are encouraged to specify their operational criteria in related studies. The lack of consensus on knowledge typologies highlights the need for new classification frameworks that incorporate the dimensions of knowledge, and the criteria identified in this study. Future research should define constituent knowledge types more clearly, supported by explicit criteria and examples. Additionally, generative AI models could be integrated with knowledge management tools designed for specific knowledge types to enhance the accuracy and applicability of their outputs.