CHALLENGE - A haptic device is an actuated human machine interaction (HMI) device that provides force and torque feedback to the operator through human sense of touch according to the reflecting force and torque from objects in a real, teleoperated, or virtual environment. It is a complex task to develop and optimize a high-performing haptic device, mainly because of the multi-domain and multi-criteria performance requirements for such devices. The general goal for the research presented in this paper is to further develop the previously proposed model-based framework and methodology into a situated and computationally efficient design framework for multiobjective optimization of haptic devices.
SOLUTION - The optimization of the 6 DOF TAU haptic device is illustrated to verify the proposed framework. The objectives for the optimization are derived from the stated requirements on the haptic device. Some of the main system requirements are listed below and a more complete set of requirements can be found in: a minimum translational workspace of 50x50x50[mm]; sufficient dexterous workspace; no singularities within the workspace; high isotropy and transparency within workspace.
From these requirements, the optimization problem can be formulated with specific objective functions, design variables and constraints. As a simplified case study, two objectives, high isotropy and large dexterous workspace, are considered. Furthermore, five continuous design variables with specified range are used and the rest of the system (such as the torque transmission chain, joints, and motor supports) are treated as model parameters. As a means to reduce the number of design variables and hence also the complexity of the problem, a DOE study is suggested as a first step in the proposed framework. The performance objective is to maximize the dexterous workspace (VI) and the global isotropy (GII), with singularity and required workspace as constraints.
Due to the non-linear and multi-objective character of the actual haptic device, a general optimization process based on the Multi-Objective Genetic Algorithm (MOGA) is used. The optimization process is implemented with the modeFRONTIER, which offer tool integration and optimization process automation. The best-fit metamodels for both objectives found previously are used in this optimization process.
BENEFITS - To use DOE as a first step in the process can definitely give insights in how to reduce the complexity of the problem; Combining DOE and metamodeling techniques in the optimization process enabled reduction of the number of design variables and their ranges and improved the optimization efficiency.