Generative design for the holistic synthesis of planar linkages, taking into account functional and non-functional requirements

• Applicants: Professor Dr.-Ing. Burkhard Corves; Professor Dr.-Ing. Arun Nagarajah
• Subject Area: Engineering Design, Machine Elements, Product Development
• Term: since 2026
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 574696223

Planar linkages are widely used in various applications, ranging from furniture to robots and manufacturing. Despite the availability of cost-effective servo motors, there remains a strong demand for linkages, especially in the textile and packaging industries, due to their ability to provide high cycle rates and energy-efficient solutions. In mechanisms synthesis, linkages are designed to meet the specific requirements of a process or technical system. Traditionally, this design process is divided into two steps: type synthesis and dimensional synthesis. Both synthesis steps pose individual challenges and require the combination of coordinated synthesis and optimization methods, as well as extensive expertise in mechanism science. However, this specialized knowledge is increasingly unavailable due to a shortage of skilled professionals. This lack of expertise often prevents engineers from fully understanding all constraints on the mechanism, especially those that arise implicitly from the given requirements. Consequently, certain predefined requirements can significantly restrict the solution space for type synthesis in advance. To address continuously evolving and volatile requirements, agile and rapid development processes are essential. The traditional two-step mechanism synthesis, without considering both explicit and implicit requirements in their entirety, when applied to modern design tasks, inevitably leads to significant compromises in the quality of the results and considerable delays in the design process. The use of artificial intelligence (AI), particularly with the recent advances in generative AI and deep neural networks, has the potential to enable holistic synthesis of planar linkages for practical applications. This approach does not require extensive domain-specific knowledge from engineers and can operate even when the initial requirements are not fully known in their implicit entirety. The authors hypothesize that, given a comprehensive dataset of diverse mechanisms, including detailed analyses of their characteristics, the entire synthesis process can be performed by a single AI system. Furthermore, it is expected that replacing the traditional two-step synthesis process with a holistic approach will bring various advantages. The research gaps to be addressed in this project include the generation of a dataset of viable solutions that considers potentially occurring requirements, AI-based mechanism synthesis integrating type and dimensional synthesis, automated evaluation of generated mechanisms with respect to requirements, and the automated completion of given initial requirements.

• DFG Programme: Research Grants