Adaptable Origami Faces: Integrating Robotics and Computer Vision with Wet Origami Techniques to Improve Facial Reconstruction Surgery

Abstract

The treatment of burn victims is an ever-evolving topic in medical dermatology, especially in relation to how skin grafts are obtained and utilized. While there are artificial skin technologies available, using donated human skin from an uninjured part of the patient’s own body provides the highest chance of recovery. However, taking undamaged skin from the patient’s body means the donor supply is very limited; therefore, it is vital that doctors maximize the skin graft’s coverage while ensuring that it maintains effectiveness and minimizes scarring. This thesis introduces novel wet origami techniques to develop an optimal way to fold and cut a flat sheet skin graft for the face (a complex, curved shape), as well as a robotic structure that implements natural facial movements into the model. Through experimentation with different types of paper, this research finds that several thin sheets of paper with an intermediate layer of liquid adhesive is the most effective material for wet origami, lending the highest degrees of biological accuracy according to Gaussian curvature and proportion analysis. A bistable hinge fold is implemented to allow for jaw movement. Dlib 68-point face landmark detection is used to obtain facial measurements via camera capture. These patient-specific measurements are used in two ways: 1) as input for a live feedback loop that controls an origami robot, mimicking movements in real time, and 2) for generating a custom origami crease pattern that automatically adapts to fit specific individuals with unique facial features.