Design and Optimization of a Compact Metacarpophalangeal Joint Angle Sensor for Robotic Finger

Abstract

The metacarpophalangeal (MCP) joint of the dexterous robotic finger exhibits two-degree-of-freedom (2-DOF) rotation, thus conventional designs require two discrete joint angle sensors for joint position control. This multisensor configuration restricts compactness and integration in dexterous robotic finger design. By embedding a miniature permanent magnet in the joint mechanism, compact measurement of 2-DOF rotation angles is achievable. Using a single three-axis Hall-effect sensor, a mapping between 3-D magnetic flux distribution and the joint’s 2-DOF angles is established. To realize high-sensitivity response, sensor’s geometry parameters were optimized. Subsequently, an artificial neural network was trained to perform inverse mapping from the measured 3-D magnetic flux intensity back to the joint angles. Experimental validation confirms that this integrated sensor achieves a maximum absolute joint angle sensing error of 0.63°. This method offers a promising, compact solution for high-precision MCP joint angle sensing.