| Summary: | Thesis (Ph. D.)--Massachusetts Institute of Technology, Dept. of Mechanical Engineering, 2002. === Includes bibliographical references (leaves 142-148). === A new type of wearable sensor for detecting fingertip touch force and finger posture is presented. Unlike traditional electronic gloves, in which sensors are embedded along the finger and on the fingerpads, this new device does not constrict finger motion and allows the fingers to directly contact the environment without obstructing the human's natural haptic senses. The fingertip touch force and finger posture are detected by measuring changes in the coloration of the fingernail; hence, the sensor is mounted on the fingernail and does not interfere with bending or touching actions. Specifically, the fingernail is instrumented with miniature light emitting diodes (LEDs) and photodetectors in order to measure changes in the reflection intensity when the fingertip is pressed against a surface or when the finger is bent. The changes in intensity are then used to determine changes in the blood volume under the fingernail, a technique termed "reflectance photoplethysmography." By arranging the LEDs and photodetectors in a spatial array, the two-dimensional pattern of blood volume can be measured and used to predict the touch force and posture. This thesis first underscores the role of the fingernail sensor as a means of indirectly detecting fingertip touch force and finger posture by measuring the internal state of the finger. Desired functionality and principles of photoplethysmography are used to create a set of design goals and guidelines for such a sensor. === (cont.) A working miniaturized prototype nail sensor is designed, built, tested, and analyzed. Based on fingertip anatomy and photographic evidence, mechanical and hemodynamic models are created in order to understand the mechanism of the blood volume change at multiple locations within the fingernail bed. These models are verified through experiment and simulation. Next, data-driven, mathematical models or filters are designed to comprehensively predict normal touching forces, shear touching forces, and finger bending based on readings from the sensor. A method to experimentally calibrate the filters is designed, implemented, and validated. Using these filters, the sensors are capable of predicting forces to within 0.5 N RMS error and posture angle to within 10 degrees RMS error. Performances of the filters are analyzed, compared, and used to suggest design guidelines for the next generation of sensors. Finally, applications to human-machine interface are discussed and tested, and potential impacts of this work on the fields of virtual reality and robotics are proposed. === by Stephen A. Mascaro. === Ph.D.
|
|---|
