| Touch sensing, along with hearing and vision, is an important human sensation endowed by the nature. The long standing pursuit to reproduce this amazing multi-modal sensing capability in man-made systems is challenged by the complex multi-physics nature of touch. Therefore, there is no standard modular hardware solution to realize touch sensing with a richness of information comparable to the nature sensors.;The thesis work aims to provide modular solutions to duplicate the nature's touch sensing through innovative sensor design. The design of the touch sensors is inspired by biological hair cells and is enabled by a novel fabrication / packaging method called direct Silicon-to-PCB Interconnect Assembly. High yield production of MEMS touch sensing units and 3D assembly of microchips with reliable electronic interface at enhanced mechanical robustness under large repeating loads is realized. The MEMS design of transducers and the structural design of the sensing system architecture are decoupled, allowing great design flexibility and ease of manufacturing.;Powered by this method, touch sensors for 3D interaction force and object physical property sensing are developed. Such sensors can instrument discrete nodes of artificial intelligent systems, such as industrial robots or MIS tools.;The framework of 3D touch sensing is expanded to a large planar surface enabled by novel mechanical design. A force proportional touch pad is developed for sensing 3D contact force, finger position and common multi-finger gestures during touch interaction, promising innovative applications in HCI. With the same platform, we addressed the key challenges faced in medical palpation training. A smart training model is constructed to record tactile data of the palpation process with feature extraction algorithms to interpret the pressure, exploratory maneuver and search pattern, the three key quantities to evaluate palpation performance.;This platform is also used to study human active tactile exploration. It is the first time that complete force profile during human active tactile exploration is recorded. Experimental design rules and interesting facts about human active exploration behaviors are discovered and analyzed numerically and analytically. The findings are valuable assets for future study of the human active touch process. |
