A Dynamic Time Warping algorithm for industrial robot motion analysis

Abstract

Industrial robots have a wide field of application and their applicability is constantly examined in further new application areas. Therefore the performance characteristics of industrial robots are of great interest. Such performance criteria are defined by national and international industrial standards, which also describe methods to evaluate them. However applying these criteria and methods directly to any robotic application may result in a false identification of robotic accuracy and consequently to a possible misinterpretation of the robots' capabilities. This paper describes the problems of using industrial standards for robot motion path analysis. Relating to the underlying motion data, similarities to signal processing methods in speech recognition are discussed and the usage of Dynamic Time Warping (DTW) as a general method for industrial robot motion analysis is proposed. We present a new variant of the DTW algorithm which allows the mapping of interpolated trajectory points without significantly increasing the computation complexity.