Task-Specific Morphological Design Optimization of Kinematically Redundant Manipulators.

摘要

Kinematically redundant manipulators are coveted for their ability to achieve not only the primary manipulation goal of maintaining kinematic dexterity, but also secondary manipulation goals such as collision avoidance and dynamic energy minimization, which afford the performance of more complex and greater varieties of tasks than possible with standard, non-redundant manipulation systems. A critical challenge in redundant manipulation is designing manipulator morphologies that maximize the achievement of secondary goals, also known as redundancy resolution, without adversely affecting kinematic dexterity. Current manipulator design fitness metrics are adequate for dexterity measurement, but they neither consider redundancy resolution nor elucidate its dependence on morphological design. This Ph.D. thesis develops novel design fitness metrics that quantify the capacity of a manipulator morphology to promote redundancy resolution over a specific set of tasks without diminishing kinematic dexterity. These fitness metrics serve as the basis of a design methodology, also proposed in this thesis, which improves redundant manipulator performance through the optimization of morphological design parameters. The efficacy of the proposed metrics is demonstrated by improving the performance of redundant manipulators used in manufacturing, commercial, and medical applications.