This paper proposes a novel learning method for prosthetic hand control. Co
nventional works have used off-line learning methods for control, and hence
two kinds of training must be carried out separately: one is for the amput
ee to control prosthetic hands, and the other is for the prosthetic hand co
ntroller to adapt to the amputee's variations. Consequently, an amputee can
not acquire the sensations of operating prosthetic hands through training,
and nevertheless he or she is likely to experience difficulties in forcing
the prosthetic hand controller to follow the change of his or her own chara
cteristics in practical use. We accordingly design an on-line learning mech
anism which can adapt to the individual's characteristics in real time. Usi
ng this mechanism, a suitable mapping function of the surface electromyogra
m (EMG) to motions of prosthetic hands can be acquired according to the amp
utee's evaluation in practical use. Thereby, the mechanism realizes a short
ening of training time and adaptation to individual variation in real time.
The experiments succeeded in discriminating six forearm motions to verify
the proposed method. First, we use intrinsic exercise images to control a p
rosthetic hand, and compare our on-line method with one conventional off-li
ne method. Second, we use EMG signals on shoulder girdles to control the pr
osthetic hand for upper elbow amputation. The discrimination rate in forear
m EMG experiments is 89.9% by our method and 80.3% by the conventional meth
od. Moreover, we show the possibility of applying the on-line learning meth
od to upper elbow amputees, because a discrimination rate of 79.3% is achie
ved by our method in shoulder girdle EMG classification. (C) 2001 Scripta T
echnica.