Deep Learning has been gaining popularity due to its numerous implementations and continuous growing capabilities, including the prosthetics industry which has trend of evaluation towards the smart operational decision. The aim of this study is to develop a reliable decision-making system for prosthetic hands which is responsible to grasp or point an object located in the interaction area. In order to achieve this goal, we have exploited the measurements taken from a low-cost inertial measurement unit (IMU) and proposed a convolutional neural network-based decision-making system, which utilizes 9 distinct measurement variables as input, 3 axis accelerometer, 3 axis gyroscope and 3 axis magnetometer. The given experiments on this paper successfully identify that the deep learning approach produces reliably 99.2% accuracy on deciding the final prosthetic hand action by analyzing IMU sensor readings that represents motion trajectory of the hand.
Cihan Uyanik, Syed F. Hussaini, Erdem Erdemir, Erkan Kaplanoglu, Ali Sekmen, "A deep learning approach for final grasping state determination from motion trajectory of a prosthetic hand", Procedia Computer Science, Volume 158, 2019, Pages 19-26, ISSN 1877-0509, https://doi.org/10.1016/j.procs.2019.09.023.