The development and testing of a glove that conveys the position of each finger of the wearer’s hand are described. Transducers are sewn into the glove at two of the three joints of each finger and the thumb. There are two such transducers at the bottom joint between the thumb and index finger, which measure the abduction. When the wearer bends a finger, the appropriate transducer produces a voltage that is proportional to the angle of rotation at the joints involved.
Software activated by the application can sample the voltage at each joint, and thereby create a digital picture of the current status of the wearer’s hand.
The authors have not stated the practical objective of their research. However, it should be clear that, at the low end of the spectrum, the wearer could transmit one of a number of answers to a question by moving one or more fingers to a particular position. In the middle of the spectrum, the user could transmit letters of the manual alphabet by positions of her or his fingers. This might require an additional transducer for wrist rotation. Actual sign language requires gestures that use the arms as well as the fingers and hands. This could be achieved with a number of additional transducers.
A number of conditions were examined with regard to the setup. First, a subject was asked to hold his hand still for a number of seconds to see if the resulting noise was objectionable; it was acceptable. Second, a number of subjects were examined to see if the glove could adjust to various hand sizes. Also, the sex of the subjects was considered. Third, the glove was constrained in a gig, and the transducers were checked one at a time to see the range of values that could be detected for that transducer while the others were held in position. Fourth, the relation between the angle of motion at the joint and the signal obtained from the glove was studied. For many of the joints, this turned out to be a fairly linear relationship. For some, however, such as that between the thumb and the index finger, a better fit was provided with a power function.
Additional calibration was performed to provide a fairly accurate correlation between signal output and joint rotation. As might be expected, it was tricky to keep track of the thumb motion because of its greater degree of freedom in three dimensions.