Digital Materials
Introduction
Digital materials are repeating assemblies made of discrete units. Like LEGOs, these units may have different material properties, but have similar geometry and reversibly fasten together. An assembly of units creates a structure where each unit placed has a limited set of positions and orientations. These structures are digital because each position on the structure is either wholly populated with a part or it isn't, and error-correction occurs during alignment to the structure.
The repeating nature of the structure lends itself to cooperative robotic assembly. Assembler robots can be small compared to the structures that they make and can be specifically designed for a particular lattice geometry. Robots such as these don't require advanced sensing systems, unlike traditional robots in unstructured environments, because they can move in whole unit steps and use the structure for realignment; they are digital and less susceptible to errors in the same way that digital signals are less susceptible to electrical noise than analog signals.
Automated assembly of digital materials is an exciting manufacturing technology. The discrete units could be flat packed for storage and transport, be assembled into arbitrary shapes and sizes by a swarm of assembler robots, and be easily recycled after use since the assembly process is reversible. If the size of the discrete unit shrinks over time, this method of manufacturing could potentially be used for making almost anything.
I have some things to figure out:
• What would these structures look like?
• How would the units be fastened together?
• How would the robots actually perform assembly processes?
Structures
I started investigating lattice structures as part of an independent study I completed in the fall of 2017 with Prof. Michael Cottom, the Technical Director of the Scene Shop in the Theater Department at UMass Amherst. I explored lightweight, strong, repeating structures that are reusable and easy to disassemble in the context of theater scenery. I focused on lattice designs that have octahedral symmetry and take up significantly less space when disassembled.
See my technical journal for more information.
This structure is made of popsicle sticks and hot glue. I wanted to make the structure from pieces of the same length so that assembly could be simplified. I learned a lot about sphere packing geometries from this structure. The popsicle sticks were not ideal so I acquired toothpicks.
This structure is made of toothpicks and hot glue. I made this pyramid large enough to have a node in the center that is connected to toothpicks in each direction. I used the geometry of this node to model a connector part so that I could do away with the hot glue. This way I could build structures faster and also disassemble them.
This structure is made of toothpicks and 3d printed connectors. The connectors enabled me to quickly play around with other arrangements using the same node angles. I liked the fact that each toothpick is the same size, but I didn't like how there were two different parts: toothpicks and connectors.
This was my first attempt at a structure made by repeating a single part. It is a classic Kelvin Lattice. There is no glue; the identical parts snap together. I designed the part to be able to be laser cut out of sheet stock for easy fabrication. I noticed that assembly of this structure required bending the pieces in order to fit the next piece in place. This bending was a complicated maneuver which required good hand-eye coordination and would not be able to be done with a simple robot.
This structure is made of paper. It is also made by repeating a single part. It is heavily based on this patent by Kenneth Cheung and Neil Gershenfeld. By making this structure, I learned that geometry has a significant effect on strength. The paper structure was surprisingly rigid. I also noticed that the assembly of this structure had the same issue as the last: it required bending the pieces to put them in place, which often resulted in creased parts.
With this structure, I went back to the toothpick set and noticed that I could create the same geometry as the paper structure. I liked this structure because the base units are pronounced and there are many degrees of symmetry; there isn't a defined top or bottom. My next objective was to figure out a way to build a structure like this without the need to bend parts during assembly. This would make automating the assembly process far easier.
Fasteners
As the lattice assembly gets larger, it becomes more rigid and difficult to bend the parts to fit the next piece into place. I needed a way to assemble the base units together without having to bend the assembly. I settled on an assembly method consisting of two steps: placing the part and then fastening it. It is critical that the fastener mechanism stays out of the way during the placement procedure, otherwise the parts will collide and won't be able to be placed in the correct location. I decided that the solution was a genderless spring-loaded captive fastener. By pulling the fingers of two adjacently placed fasteners together and twisting, they will mate together with an audible click. When unmated, the fingers automatically retract via the spring action, so as not to get in the way of assembly or disassembly. There is also a wave pattern on the fasteners' mating surface to ensure that the base units are rotationally aligned.
Here is a single octahedron base unit from the last structure of the previous section, which effectively consists of 8 of these units. I challenged myself to come up with a way to fasten these units together without increasing the number of parts or reducing symmetry. Each connector half would fasten to an identical connector half.
This was quite a challenge. After several revisions, I came up with this design, which achieves what I set out to do. It ended up being a fairly complicated part which I scaled up 3 times in order to get the 3d printing resolution I needed. I considered this a success but was unhappy with the size of the connector. My next step is to find a way to fabricate this part in a smaller size so it is more reasonable to work with and mass produce.
Manual Assembly
In the spring of 2019, I took a course called Material Experiments in Landscape Architecture taught by Prof. Carolina Aragón. My research paper in this course was focused on digital materials, and my project was to scale up the paper structure I made previously to be larger than myself (see picture below). It is a structure made of roughly 1600 identical X-shaped laser-cut parts. My term paper includes the research and project documentation.
_JPG.jpg)
Robotic Assembly
Ongoing research has solved many components of robotic assembly of digital materials, however, I have yet to see a demonstration where a relative robot can reversibly fasten parts which are flattenable, passive, and discrete onto arbitrary locations on the grid on the surface of a cellular lattice structure. This is what I intend to demonstrate. Most existing robotic assemblers described in the literature are either theoretical and somewhat vague, considerably constrained in the geometry of the structures that they can produce, use actively powered or unflattenable discrete units, or use magnets for attracting the discrete units together, which easily come apart when the structure is loaded, since they are not mechanically fastened.
​
I plan to have self-reconfiguring modular robots pick up, transport, and place each part on the cellular grid on the surface of the structure, while a separate robot will locomote within the structure and be able to perform the fastening procedure on the genderless captive fasteners, which are components of each discrete part. These will be a more mature variation of the fastener shown in the fasteners section and will eliminate the challenges of alignment and dexterity that come with typical strut and node assemblies. The fastening procedure needs to be done from within the structure because some of the fastened joints are often inaccessible from the surface once parts are placed.
​
See my technical journal for more information.
This preliminary animation shows the capabilities I am striving to implement. Blue robots locomote on the surface of the structure, connect and disconnect from each other, and pick up and place blocks. Green robots locomote within the structure and fasten and unfasten passive blocks. Details will be refined next.
This animation shows the internal detail of a single cellular block and how its fasteners function. The fastener's complicated internal geometry is necessary to fulfill the design requirements. It needs to reversibly fasten to 4 struts and an identical fastener without getting in the way of passing robots or blocks.
The fasteners are 3D printed with the Ultimaker S5. This video shows the S5 printing a batch of 87 fasteners. The S5 can print with both build material and soluble support material with high enough resolution to fabricate the fasteners 25mm wide without significant defects. The S5 enabled printing at this small scale.
This video shows the manual assembly process for one block. The 6 fasteners were 3D printed and the 12 struts are cut from pultruded carbon fiber. A tool similar to a screwdriver is used to actuate the fastening mechanism so that the struts do not fall out. This assembly process will become automated in the future.
This video shows the difference in stiffness of blocks made with struts of different materials. Shown are struts made of flexible TPU, traditional PLA plastic, and rigid pultruded carbon fiber. Selectively placing flexible and rigid blocks in a structure opens possibilities for additional functionality.
This video shows a rudimentary compression test of a block made with carbon fiber struts and PLA fasteners. It failed at around 90 pounds. The two failure modes for these tests were shearing off the pin of the fastener keeping the strut in place and blowing out the back of the pultruded carbon fiber strut.
This video shows the manual assembly process to fasten two blocks together. Wrench tools are used to grab and twist the fastening mechanisms into a locked position. This assembly process is reversible and will be automated with the green robots shown in the first clip of this section.
This rough screen capture shows mock-up robots with more detail than the first animation. The robots unfasten a block and move it away. These mock-up robots don't have motors in their design yet. The (green) robot's end effectors grip and engage the fastening mechanism with two pairs of 4 gears.
This video shows the mock-up robots from the last video in 3D printed form performing the same process. They work as expected. The next step is to add motors. Some of the blocks shown here have laser cut acrylic struts, which are very easy to make but are less rigid than the pultruded carbon fiber struts.