Functional Principal Component Analysis for Recognition of Arm Gestures and Humanoid Imitation

Abstract

This paper investigates the use of functional principal component analysis (FPCA) for automatic recognition of dynamic human arm gestures and robot imitation. FPCA is a statistical technique of functional data analysis that generalizes standard multivariate principal component analysis. Functional data analysis signals (e.g., gestures) are functions that are considered as observations of a random variable on a functional space. In particular, FPCA reduces the dimensionality of the input data by projecting them onto a¯nite-dimensional space spanned by a few prominent eigenfunctions. The main contribution of this work is the proposal of a novel technique for unsupervised clustering of training data and dynamic gesture recognition based on FPCA. FPCA has not been considered in previous studies on humanoid learning. The proposed approach has been evaluated in two experimental settings for motion capture. In the¯rst setup single arm gestures are recognized from inertial sensors attached to the arm of the user. In the second setup the method is extended to two-arm gestures acquired from a range sensor. Recognized gestures are reproduced by a small humanoid robot. The FPCA method has also been compared to a high performance algorithm for gesture classi¯cation based on dynamic time warping (DTW). The FPCA algorithm achieves comparable results in both recognition rate and robustness to missing data, while it outperforms DTW in terms of e±ciency in execution time.

Recognition of a Single Dynamic Gesture with the Segmentation Technique HS-ab and Principle Components Analysis (PCA)

“…In [ 25 ], PCA characterizes the function of the fingers in a vision-based Chinese SLR system. Aleotti [ 26 ] analyzed the use of PCA for automatic recognition of dynamic gestures of the human arm and the imitation of the robots. In [ 27 ], PCA is applied for the recognition of hand gestures in order to represent the alphabet.…”

Inverse kinematics of a NAO humanoid robot using kinect to track and imitate human motion

“…Benbrahim and Franklin [7] implemented balancing of biped robot using this methodology. Another method for imitation was proposed by Aleotti et al [9] and Aleotti and Caselli [10]. They developed a statistical gesture recognition and imitation system based on Functional Principal Component Analysis (FPCA).…”

Converting constrained whole-body human motions to humanoid using smoothing

“…This smoothing process has two advantages: first, when we try to expand the framework of motion generation, we can classify different motions, tuning or blending motions in the lower dimensional space, for instance by using Functional PCA which is a statistical method. J. Aleotti et al used Functional PCA and classified different motions in lower dimension and obtain representative motions to input the humanoid robot [1]. However, they applied the method to upper body motion, but not to whole body, without considering stability conditions.…”