Character Recognition on a Humanoid Robotic Platform via a Laban Movement Analysis

Abstract

An active area of research is exploring how to mimic human movement on a robotic platform. One step toward achieving this goal is the ability to create a method in which a robotic platform successfully portrays a character or character traits that a human can recognize. Ballet is a performing art in which recognizable character types are conveyed through movement. This paper explores the recognition and differentiation of archetypal characters used in classical ballet based on their upper body movements and applies this information to a robotic platform. An observational guide was created using Laban Movement Analysis (LMA) for analyzing the movements of distinct characters found across several classical ballets. Eleven examples were chosen: three villain character types, three bird character types, and five dying character types. The upper body movements of these characters were tracked utilizing the observational guide to see if their movements utilized the LMA characteristics in distinguishable ways. The results from tracking the movements were analyzed, and movement sequences were created on a NAO robot to emulate these character types: a process subsequently validated by a user study.