FusionTRIZ Framework for Cross-Disciplinary Innovation: A Neuromorphic Imaging Application

Innovation often depends on transferring ideas across disciplines, yet finding relevant knowledge outside one’s field remains difficult and time-consuming. Conventional search tools typically emphasize keyword, semantic, or citation similarity rather than the deeper analogical or interdisciplinary connections that frequently drive breakthroughs. This paper presents FusionTRIZ, a framework that combines large-scale semantic literature mining with TRIZ (Theory of Inventive Problem Solving), a codified set of 40 inventive principles distilled from decades of patent analysis. FusionTRIZ operates in four steps: (1) retrieve in-domain solution papers from a large scientific corpus; (2) use a large language model (LLM) to extract each solution’s TRIZ profile; (3) retrieve highly interdisciplinary solutions from other domains; and (4) filter cross-domain papers by their TRIZ alignment with the in-domain profile. This process surfaces conceptually relevant ideas even when the technical language, terminology, or application area differs greatly. In a focused case study (neuromorphic imaging) on tracking rapid motion in low light, FusionTRIZ identifies a polarization-enhanced sensing design from space-target tracking research. This bioinspired approach matches multiple inventive principles found in domain literature. Results suggest that combining TRIZ principle extraction with automated retrieval expands the search space for analogies, reduces ideation effort, and accelerates systematic innovation across diverse fields.