Towards generative design in the development of mechatronic integrated devices (MID)

Abstract The spatial integration of mechanical and electrical components is a key enabler for the development of innovative cyber-physical systems. These Mechatronic Integrated Devices (MID) are necessary to meet the increased demands for miniaturization and functional integration. At the same time, the development complexity of these products as well as the dependencies of the production technologies increase. Product and production process must be developed integratively from the outset. The development of MIDs is supported by cross-domain development methods (e.g. VDI2206). However, the generation of solutions is often accompanied by individual solutions such as construction tools for 3D geometry design, routing on spatial surfaces and 2D logic planning for the topology of the mechanics. On the basis of initial domain-specific solutions, a complex, highly iterative optimization process takes place, in which the MID components are adapted to geometric and production-specific conditions. Generative Design offers a promising approach for the automation of design tasks. However, this approach is used outside of architecture only in mechanical designs and does not allow the automation of design aspects of complex multi-domain components like MID. This paper presents an initial approach towards the use of Generative Design in the development of Mechatronic Integrated Devices. The focus is to develop a hybrid design systematic that allows the (partially) automated evolvement of MID components based on a product specification and solution elements to be integrated. The Generative Design process is initiated by the specification of product requirements and constraints (e.g. existing mechanical solution elements). A knowledge base formalized via ontology is used to create a principle solution. Based on an initially generated population the biological-inspired evolution is triggered. Design variants are first synthesized together with 3D routing and then selected based on fitting functions. Satisfactorily evaluated MID variants are continuously mutated and recombined until the final design is created.