Object slippage is a significant challenge toward stable grasp for prosthesis hands due to uncertainties, such as changes in object weight or fast maneuvers. Slippage can negatively affect a user’s inclination for long-term use of a prosthesis. The present study suggests a control algorithm for object slippage using surface electromyography (sEMG) and a force-sensitive resistor (FSR) mounted on the prosthesis fingertip. For this purpose, the grasp procedure is divided into four phases, labeled as open, stay open, close, and stay closed, based on the sEMG signals. In addition, a proportional-integral (PI) controller is used to adjust for the desired grasp force. The algorithm uses a decision tree (DT) classifier to determine either the occurrence of slippage or the presence of a stable grasp. A data set is experimentally collected using the FUM 3-Fingered Myo hand. Objects with increasing weight are used, and the corresponding FSR changes are recorded. Results show that the proposed DT method predicts slippage occurrence with 95.6% accuracy with an average detection time of 16.26 ms. Moreover, the proposed method can successfully anticipate slippage at least 80 ms before it actually occurs and, consequently, avoid object slippage by increasing the grasp force.

Keywords