Slippage Sensory Feedback and Nonlinear Force Control System for a Low- Cost Prosthetic Hand

The low acceptability of current prosthetic devices can be attributed to the extensive psychological effort and the high cost associated with them. To address these concerns, an on-line slippage detector was developed using only inexpensive force sensors placed at the tips of a prototype hand. The prototype consists of a five fingered prosthetic hand consisting of active digits driven independently by DC motors. Force sensor resistors (FSR) are placed at the tip of each active finger and potentiometers are attached at the proximal and middle joints. Using the information from the FSR, not only can we detect the level of normal force exerted but also slippage between the fingers and the object by calculating the fluctuations of the exerted force. An on-line algorithm is developed to calculate the derivative of the force and determine when slippage is produced. Nonlinear model predictive control (NMPC) is used to provide feedback control to the prosthetic device. It utilizes a neural network to model the dynamics of each finger. Using this model, it is possible to predict future plant performance (the amount of force exerted by the prosthetic hand). Consequently, the controller uses this prediction to calculate the best input (current needed to drive the actuators) for the system to obtain the desired output over a specific time horizon. In order to calculate the future control inputs, the optimization system minimizes the cost function associated with the difference between the measured force and the reference / target output. Experimental protocols involve grasping various objects and inducing slippage. Data was collected using the NI DAQ cards and LabVIEW software. Experiments showed promising results using this strategy in which the force exerted on an object can be modulated without additional efforts from the users.