The Challenge of Manufacturing Between Macro and Micro
Classic ways of folding paper into dynamic shapes—origami, pop-up books—inspire methods to engineer millimeter-scale machines.
Know When to Fold Them
In our laboratory, my colleagues and I saw that folding has many possible benefits over other assembly methods, such as the potential for single piece construction and the relative simplicity of 2D machining methods. In addition, folded beams—which may act as structural components in devices—use material efficiently and have high stiffness-to-weight ratios compared to solid beams. Furthermore, starting from a quasi-2D sheet trivializes the integration of electrical components as it becomes possible to use standard pick-and-place printed circuit board assembly tools.
In fact, printed circuit boards themselves are not limited to 2D structures. Thin polyimide-based flexible circuits are now ubiquitous as folded components in modern electronics. Some of these components are designed to fold just once, such as to make a board fit into an irregular shape such as in the smartphone example above. In other cases, the flexible parts can move repeatedly, such as in the read-write circuit in a disk drive. To date, these folds have been serial folds, with one step following another, like folding a paper airplane. An expansion of this to parallel folds, where a number of single folds simultaneously happen at different points on the device, is the enabling innovation for our process inspired by pop-up books. This innovation came about by necessity—researchers in my group were making small robots by folding, but performing the assembly by hand was tedious and lacked repeatability. In conversations with two of my then-graduate students, Pratheev Sreetharan and Peter Whitney, we decided that there must be a better solution—one that mirrored the complexity of the folded structures that appear in children’s pop-up books and was created with similar ease (i.e., a simple turn of the page pops everything into place).
One obvious question is how much diversity of shape we can create just by folding. Fortunately, computer scientists have proven that any polyhedra can be folded out of a single continuous sheet of paper. Furthermore, by controlling the specifications of the polyhedra, we can create nearly any linkage mechanism. The only obvious limitation in creating mechanisms purely by folding is the lack of continuously rotating joints—flexures have a finite range defined by geometry and material properties.
Origami-like techniques have been used to create intricate sculptures with thousands of folds, but they require many hours of work by hand. Similar folding mechanisms have been shown to be ubiquitous in nature, from the folds in protein structures to the unfurling of leaves and insect wings when they first emerge. There are also myriad applications in engineered systems, including parachute packing, unfolded solar panels on satellites, and quick-deployable shelters. For engineered systems, alternatives to tedious hand folding include the use of pneumatics or hydraulics to expand and shape devices, and also the use of materials with shape memory that will return to their original shape even after being confined. But part of the challenge is making it work at the scales useful to manufacturing, particularly at the meso-scale.
Another important question for assembly by folding is how to design folded structures and mechanisms. Origami has historically lived in the realm of art, and the most impressive pieces have come from a small number of highly experienced, talented individuals—not a practical source for rapid machine design. There has been progress in creating design tools that generate fold patterns automatically given a 3D geometry. But such tools are not currently at the same level as those for established procedures, such as integrated circuit fabrication.
A final potential challenge is not just in the creation of static folded structures (i.e. structures incapable of movement) but in designing articulated mechanisms that can be assembled by folding. This requires two types of folds: static assembly folds used to define geometry and structural properties, and dynamic folds that are flexures used to constrain motion along desired axes. This goal is a further challenge for design automation tools, potentially involving highly complex kinematics (and it could be even more difficult to define the inverse kinematics that would “unfold” the device into the 2D crease patterns necessary for fabrication by folding).