This thesis presents an optimization and verification of direct ink writing (DIW)-based soft sensor location for finger 3D motion measurement in the dorsum of a hand. Soft sensors were used to measure finger motions, but they had two major problems: 1) the sensors located between fingers restricted natural finger motions, and 2) the system could not measure thumb motions and abduction and adduction of other four fingers, i.e., 3D finger motions. To solve these problems, in this thesis, the soft sensors are all located in the dorsum of the hand with minimizing the number of sensors to measure the main seven finger motions. The target seven finger motions include 3 degrees of freedom (DOFs) thumb motions, 2 DOFs index finger motions, and 2 DOFs middle finger motions. As the first step, performance of the DIW-based soft sensor was verified including resolution, linearity, hysteresis and sensitivity (i.e., gauge factor). To measure target seven finger motions with less than seven sensors, an optimization algorithm of the sensor location is proposed. The cost of all sensor location combinations was calculated while increasing the number of sensors from two to six. The cost function consisted of sum of normalized mean square error and normalized maximum error with range of motion of seven finger motions. The accuracy of finger motions measurement was improved by using decoupling algorithms. The results of optimization were validated by a camera-based motion capture system. Lastly, the proposed soft sensor system was manufactured as single sensor sheet for an easily wearable system. The accuracy of system was also validated by a camera-based motion capture system.
Publisher
Ulsan National Institute of Science and Technology (UNIST)