A SYSTEMATIC METHODOLOGY TO DESIGN SOFT MACHINES BASED ON TOPOLOGY OPTIMIZATION

Soft systems, defined by their soft bodies and ability to undergo large deformations, offer advantages in adaptability, safety, and resilience compared to their rigid counterparts. In this dissertation the term soft machine is used in the title to reflect the broader category of soft-bodied systems. However, the term soft robot is employed throughout the text to remain consistent with common terminology in the field. In most soft robots, the body and actuator are inherently integrated, often forming a single deformable system. A major challenge in the design of such soft systems is the absence of a systematic, principle-based methodology that considers the predefined design requirements. Current designs are often driven by biological inspiration and personal intuition, rather than structured approaches grounded in physics and engineering. This dissertation addresses the identified knowledge gaps by proposing a top-down, principle-based design methodology for soft systems. The methodology begins by translating customer needs into a technical specification of design requirements and then proceeds through a structured sequence of concept design, embodiment design, and detailed design. It leverages common soft building blocks, such as bellows and air chambers, and incorporates a nonlinear finite element method (FEM) framework capable of creating mathematical models and capturing large deformations. Genetic algorithm, as an evolutionary algorithm, is employed to automate design exploration and optimize the topology and actuation of soft system. The developed methodology is applied to design a soft finger for a gripping application, achieving 5 cm of tip deformation using soft components. Bellows and air chambers serve as the fundamental building blocks in the design. Verification is conducted through both numerical simulations using commercial FEM software (ANSYS) and experimental testing via physical prototyping. In a verification case involving five bellows, the proposed method demonstrated small relative error compared to ANSYS simulations. In the case study, measurements from the physical prototype demonstrated strong agreement between the experimental and simulation results. The methodology also automates the identification of active versus passive actuation states, reducing the reliance on designer intuition during the concept design phase. A key conclusion is that the proposed methodology enables systematic, principle-based design of soft robots, moving beyond the bio-mimicking approaches. By representing the underlying principles of biological organisms through engineering building blocks, the methodology facilitates their integration into a general, rational, and structured design process. Furthermore, the case study confirms the feasibility, functionality, and accuracy of the proposed approach in designing real-world soft systems. This dissertation contributes to the field of soft robotics by providing a comprehensive design framework that begins with the specification of design requirements and supports a logical and methodical development process. It is also worth noting that the proposed methodology has the potential to complement and integrate existing biomimetic design approaches found in the literature.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.