LAUSR

For transradial amputees, especially those with insufficient residual muscle activity, it is challenging to quickly obtain an appropriate grasping pattern for a multigrasp prosthesis. To address this problem, this study proposed a fingertip proximity sensor and a grasping pattern prediction method base on it. Rather than exclusively utilizing the EMG of the subject for the grasping pattern recognition, the proposed method used fingertip proximity sensing to predict the appropriate grasping pattern automatically. We established a five-fingertip proximity training dataset for five common classes of grasping patterns (spherical grip, cylindrical grip, tripod pinch, lateral pinch, and hook). A neural network-based classifier was proposed and got a high accuracy (96%) within the training dataset. We assessed the combined EMG/proximity-based method (PS-EMG) on six non-disabled subjects and one transradial amputee subject while performing the “reach-and-pick up” tasks for novel objects. The assessments compared the performance of this method with the typical pure EMG methods. Results indicated that non-disabled subjects could reach the object and initiate prosthesis grasping with the desired grasping pattern on average within 1.93 s and complete the tasks 7.30% faster on average with the PS-EMG method, relative to the pattern recognition-based EMG method. And the amputee subject was, on average, 25.58% faster in completing tasks with the proposed PS-EMG method relative to the switch-based EMG method. The results showed that the proposed method allowed the user to obtain the desired grasping pattern quickly and reduced the requirement for EMG sources.