Fourier analysis of robot trajectories in random tasks

Abstract

A new method for the study and optimization of manipulator trajectories is developed. The novel feature resides on the modeling formulation. Standard system descriptions are based on a set of differential equations which, in general, require laborious computations and may be difficult to analyze. Moreover, the derived algorithms are suited to 'deterministic' tasks, such as those appearing in a repetitive work, and are not well adapted to a 'random' operation that occurs in intelligent systems interacting with a non-structured and changing environment. These facts motivate the development of alternative models based on distinct concepts. The proposed embedding of statistics and Fourier transform gives a new perspective towards the calculation and optimization of the robot trajectories in manipulating tasks.