A human-AI collaborative generative bio-inspired design framework utilizing scientific laws implicit in biological information

Bio-inspired design is a promising approach for innovation. However, the transformation of biological information into engineering solutions remains a significant challenge. To address this problem, we present a human-AI collaborative generative bio-inspired design framework that systematically extracts and applies scientific laws implicit in biological information. It begins with an AI performing ontology-constrained semantic analysis to extract physical quantities. These quantities are then used to infer scientific laws via a reasoning model. Subsequently, the laws are integrated into a visualised scientific law chain, which represents the physical logic governing biological functions. Finally, a scientific law-engineering instance matrix is utilised to prompt AI to generate a descriptive design concept, which the human designer then refines and concretises. In order to operationalise this framework, we developed a toolset that includes a physical quantity ontology repository, a scientific law repository, a scientific law extractor, and a conceptual generator. The feasibility of the framework and toolset is demonstrated through two conceptual design cases: a controllable jumping robot leg and a pipeline inspection robot. This framework provides a pathway for bio-inspired conceptual design, which enables designers to leverage the fundamental scientific laws to develop innovative solutions.