Design of pneumatic and hydraulic soft actuators by topology optimization method.

• Ano de defesa: 2020
• Autor(a) principal: Souza, Eduardo Moscatelli de
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Biblioteca Digitais de Teses e Dissertações da USP
• Palavras-chave em Português: Atuadores moles; Carregamento dependente do domínio; Compliant mechanisms; Design-dependent loads; Mecanismos flexíveis; Método dos elementos finitos; Otimização topológica; Robôs; Soft robots; Topology optimization
• Link de acesso: https://www.teses.usp.br/teses/disponiveis/3/3152/tde-21102020-111748/

Soft robots are machines completely or partly designed from compliant materials (such as elastomers) to circumvent limitations presented by stiff robots like difficulties to operate in unstructured environments and to handle fragile objects. In general, soft robots are capable of realizing complex movements by the deformation of their own structure. Therefore, it is necessary to find an appropriate structure for each movement requirement. This can be facilitated by automation techniques, so, recently, researchers are investigating automated design techniques to create enhanced soft actuators. In this context, this dissertation is dedicated to study the application of density-based topology optimization to the design of soft actuators driven by pressure loads (pneumatic and hydraulic). The approach followed is to synthesize compliant mechanisms actuated by design dependent loads. The compliant mechanisms are synthesized by selecting the maximization of output displacement as objective function and the design dependent load problem is solved by using mixed displacement-pressure finite elements. This work shows that undesired open designs may be obtained if no constraint is used, becauseholes may appear during optimization. Therefore, a projection technique is proposed to avoid open designs and its effectiveness is demonstrated by the solution of three numerical examples: a bending, a linear, and an inverter actuator. The finite element problem is solved by using FEniCS framework, the optimization is carried with an interior point algorithm (IPOPT), and part of the sensitivity analysis is automated by dolfin-adjoint package.