Research On Mechanical Design And Bionic Control Of Multifunctional Prosthetic Hands

A survey conducted in 2010 reported about 247 thousand physically disabled persons in China and this number will mostly likely increase in the future. Physically disabled persons such as the upper limb amputees are mostly restricted from performing different tasks that require their arm functions. Therefore, developing an intelligent bionic prosthetic hand would help this category of amputees to carry out their desired hand functions during activities of daily living(ADLs). However, most existing commercial prosthetic hands are expensive, relatively heavy, and impractical, which makes them relatively difficult for amputees to use. The state of the art method for prosthetic hand control is based on Electromyography Pattern Recognition(EMG-PR), that decodes the hand motion intentions of amputee from their stump arm EMG signals. However, performance of EMG-PR based prosthetic hands are not yet available for clinical use and one reason might be that their classification performance is not sufficient enough in clinical setting.Based on an extensive analysis of the frequently observed hand movements in ADLs, this paper presents the design of a low-cost prosthetic hand that incorporates the all the commonly observed hand motions. The costs of this prosthetic hand is basically minimized by reducing the number of actuators. In addition, a 3D printing technology was used to build a prototype of the designed prosthetic hand. Furthermore, the newly designed prosthetic hand was evaluated and results reveals that it could reliably achieve 85% of hand motion during ADLs. And its crawl coverage and gripping force has been tested in an experiment. The results imply that this prosthetic hand could help amputees achieve most of the hand movements required during ADLs.The majority vote technique was applied to the results of the EMG-PR to enhance the performance of the prosthetic hand controller. And It was realized that some erroneous patterns that could compromise the classification performance of the EMGPR controller were accurately rectified. The performance of the majority vote technique was evaluated by using the dataset of seven healthy subjects. From the obtained results, we realized that majority vote technique significantly improved the performance of the prosthetic hand controller.