Deformable Media for Visual and Tactile Interfaces

摘要

We experience a variety of natural touch surfaces in our daily lives. These surfaces range in compliance from hard to soft, and in texture from smooth to rough. Human computer interfaces, on the other hand, have largely been of the rigid type; surfaces that don't deform under the forces of touch. In the last 15 years, we have experienced technological paradigm shifts (e.g., VR, wearable computing) that have introduced a need for a more capable physical interface. Soft computer interfaces offer another dimension to touch interaction, and a medium with which to create better abstractions of natural surfaces. This dissertation explores the use of soft membranes as a medium for human computer interaction. Specifically, I address three questions: (i) how do we incorporate visual display into a soft haptic interface, (ii) how do we recognize human touch in a deforming medium, and (iii) how ought we quantify information in a deforming medium? I address these questions through three threads. First, I present stretchable displays, based on hyperelastic light-emitting capacitors (HLEC's), that can be embedded in an elastomer to actively display information under large deformations. The HLEC system stretches to >500% strain in uniaxial tension, which as of this writing, exceeds other systems by >4X. Secondly, I present OrbTouch, a soft touch interface that interprets human touch. This system uses embedded arrays of capacitance sensors in combination with a convolutional neural network-based signal processing layer to learn touch patterns from human users. Finally, I provide information theoretic arguments that relate information produced in capacitance signals to the underlying deformations that cause them.