TOWARDS A METHODOLOGY FOR MULTIDISCIPLINARY DESIGN OPTIMIZATION

Author(s): 
Xuan Sun, Kjell Andersson, Ulf Sellgren (KTH Royal Institute of Technology)

CHALLENGE - Design of haptic devices requires trade-off between many conflicting requirements, such as high stiffness, large workspace, small inertia, and others. With the traditional design and optimization process, it is difficult to effectively fulfill the system requirements by separately treating the different discipline domains. To solve this problem and to avoid sub-optimization, this work proposes a design methodology, based on Multidisciplinary Design Optimization (MDO) methods and tools, for optimization of six degree-of-freedom (DOF) haptic devices for medical applications, e.g. simulators for surgeon and dentist training or for remote surgery. 

SOLUTION - Because of the high-level requirements on haptic devices for medical applications in combination with a complex structure, models such as CAD (Computer Aided Design), CAE (Computer Aided Engineering), and kinematic models are integrated in the optimization process and present a systems view to the design engineers. An integration tool for MDO is used as the framework to manage, integrate, and execute the optimization process. A case study of a 6-DOF haptic device based on a TAU structure to be used as a surgical simulator for training on hard tissues, such as bones and teeth is used to illustrate the proposed methodology. With this specific case, a Multi-objective Genetic Algorithm (MOGA) in modeFRONTIER is applied, with an initial population based on a pseudo random SOBOL sequence and Monte Carlo samplings is used for the optimization. The performance objective is to maximize the dexterous workspace (VI), and the global isotropy (GII), while minimizing the total mass (MI) of the device, with singularity and required workspace volume as constraints. 

BENEFITS - The optimization results for this case study of the 6-DOF TAU haptic device is shown in Figure 10 above. Since the optimization is formulated with the target to minimize all three objectives, a smaller value in the result indicates a better design. 

LOG IN TO DOWNLOAD