Design of serial link manipulator with uncertainties using interval method

Abstract

This article proposes a performance-measure based design of a serial link robot manipulator, addressing the effects of dimensional uncertainties on the performance so that a specific task can be accomplished. The design problem is formulated as a constrained optimization problem, where the design variables and constraints are defined in intervals. The objective function being optimized, incorporates some chosen manipulator performance measures (obtained from literature) which include global manipulability, global conditioning index and dynamic manipulability, such that both geometric and inertial properties are covered and the solution will provide not only nominal values of the design parameters, but also a kind of tolerance, ensuring the values of the design variables lying within a range. This co-design of nominal values and its tolerance is obtained by casting the whole problem using interval techniques. The design problem is solved in the twostep process. Firstly, link lengths are determined through maximizing global manipulability and global conditioning index and secondly, inertial properties, in terms of crosssectional geometry are determined through maximizing global dynamic manipulability. To show the effectiveness of the proposed method an exemplar problem of the design of twolink two degrees-of-freedom planar manipulator under gravity and with payload is considered.