It is important to choose the parameters of a robot manipulator (configuration, dimension, motors, etc.) that are most suitable for the required robot tasks. Considerable research has been done in this area. Depkovich and Stoughton [11] proposed a general approach for the specification, design and validation of manipulators. The concept of Reconfigurable Modular Manipulator System (RMMS) was proposed by Khosla, Kanade, Hoffman, Schmitz, and Delouis [24] at Carnegie Mellon University. There goal is to create a complete manipulator system, including mechanical and control hardware, and control algorithms that are automatically and easily reconfigured.
Designing an optimal manipulator is not yet well defined, and it depends on the definition and criterion of optimality. There are several techniques and methodologies to formalize this optimization problem by creating some objective functions that satisfy certain criteria, and solving these functions with the existence of some constraints.
One criterion that is used is a kinematic criterion for the design evaluation of manipulators by establishing quantitative kinematic distinction among a set of designs [6, 40, 41]. Another criterion is to achieve optimal dynamic performance; that is to select the link lengths and actuator sizes for minimum time motions along specified trajectory [38, 48].
TOCARD (Total Computer-Aided Design System of Robot Manipulators) is a system designed by Takano, Masaki, and Sasaki [51] to design both fundamental structure (degrees of freedom, arm length, etc.), and inner structure (arm size, motor allocation, motor power, etc). They describe the problem as follows: there is a set of design parameters, a set of objective functions, and a set of Gavin data (constraints). The design parameters are:
The objective functions for the design of robot arm are as follows:
The constraints can be:
Hollerbach proposed an optimum kinematic design for a seven-degree of freedom manipulator [19].