Method for Automatic Slippage Detection with Tactile Sensors Embedded in Prosthetic Hands

Tactile sensing is fundamental for the human hand to achieve high dexterity. Most prosthetic hands are still devoid of tactile sensors, implying that the user cannot perceive external stimulation nor react in a fine manner. As a consequence, unforeseen events, e.g. slippage, are difficult to manage. This paper proposes an algorithm to perform slippage detection with tactile sensors integrated into prosthetic hands. The algorithm is based on the filtering of the tactile sensor output; rectification and envelope follow the filtering. A binary signal, relating to slippage, is finally computed. An electrical circuit has been designed and built to elaborate the tactile signals. These have been embedded in a bioinspired fingertip mounted on a prosthetic hand, which has been interfaced with a robotic arm to assess the algorithm capability to identify slippage. Eight different surfaces have been employed, while three sliding velocities have been tested with a random interaction force between the fingertip and the test surfaces. Finally, experiments in a closed-loop configuration have been conducted to demonstrate the algorithm effectiveness in dynamic manipulation. Results proved the adequacy of the algorithm in terms of slippage detection and short latency between onset of slippage, actual detection and hand reaction.