Analysis and evaluation of hand motion in use of tools

The purpose of this paper is to discuss how to evaluate quantitatively the complicated human hand motions in use of tools. The targets of this evaluation method are cyclical hand motion in use of tools. The advantage of this evaluation method is it is adptive for any hand motions without dependent on tools used. The hand motion is evaluated by comparing its characteristic with expert´s one in terms of ´Accuracy´, ´Repeatability´ and ´Efficiency´. The characteristics of the hand motion are obtained based on the statistical method. The characteristics are one-dimensional data gathered with Principal Component Analysis (PCA) and quantity of the movement. The angle data measured by CyberGlove are classified with k-mean method by their pattern, phase-lag and quantity of movement. The effectiveness of the proposed evaluation is shown through the experiments of two hand motions. One of them is the hand motion in use of chopsticks. As a result of this experiments, the chracteristics of this evaluation method is shown. Another of them is the hand motion in revolving two balls alternately on the palm. As a result of this experiments, the change of evaluation values in learning process is shown. Due to evaluate only proficiency of hand motions, this method failed to explain weak points of subject´s hand motions. The characteristics, which were provided through the experiments of using this evaluation method, are useful for developing a instruction system of the hand motion in the future.