Phantom | An Bio-inspired Interactive Prototype
Final Project for Design & Making Across Disciplines Studio | Cornell Tech & Cornell AAP College
[DATE] Sept - Dec 2022
[INSTRUCTOR] Jenny Sabin (Architecture)
[Participating Instructors] Jonathan Butcher (Biomedical Engineering), Nate Cira (Biomedical Engineering), Wendy Ju (Information Science), Marty Murtagh (Materials Science & Engineering), Uli Wiesner (Materials Science & Engineering)
[Teaching Associate] Nick Cassab
[Design Team] Chenming He, Ziqi Wang, Minyu Huang
[My Role] Concept Design, Algorithm Research, Digital Tool Development, 3D printing, Digital Modeling, Physical Models, Video Making
[Abstract] Inspired by the Differential Growth phenomenon in nature and its underlying morphogenesis mechanism, this project aims to introduce a new interactive prototype achieved by the combination of generative design, customized 3D printing on elastic fabrics, and the Arduino System. This prototype can be further developed to be implied in designing an interactive installation, dynamic architecture facades, wearable electric clothing, soft robots, and other fields.
During this exploration process, we built up the algorithm to mimic differential growth process and created our own digital tool to generate complex printable geometry. Then we leveraged customized 3D printing on the elastic fabric to enable the fabric to have a differential deformation ratio in different parts, which allow the fabric to deform naturally like leaves in nature. Next, the mechanical system and our delicately designed interactive light system are added to the system to facilitate to complete the basic interactive prototype, in which the form of fabrics and lighting effects on them can change according to human behaviors or other external stimulus responsively. Lastly, based on this prototype, we tried to design a large-scale interactive installation named Phantom over an urban public plaza as an example to showcase how this prototype can contribute to an installation design to make the urban environment more lively and increase public engagement.
Stage 1. Differential Growth Algorithm Exploration
In recent years, designers and scholars have been trying to get some inspiration from the biology field and then apply it to the design field to get more possibilities and varieties. Among those attempts, Differential Growth, a common natural phenomenon, just as its literal meaning, refers that living creatures have different growth ratios in different parts, has drawn a lot of attention. Since lots of complex behaviors, like phototaxis, or patterns, such as the unique texture of Brian Corals, can all be explained or described as the result of Differential Growth, lots of precedent researchers have devoted to decoding this phenomenon into the digital algorithm and then tried to use it to reproduce the natural complexity.
After reading lots of related scientific papers in biology and morphogenesis field, I learned that every single complex natural pattern or behavior is caused by cells’ uneven proliferation and movement. So that if we can simulate cells’ non-uniform distribution and growth, then we can probably reproduce the complex macroscopic phenomenon. Based on this thought and also inspired by lots of previous practice experience of digitalizing Differential Growth phenomenon from other computational designers, I finally figured out that Differential Growth could be simplified and decoded into an algorithm based on a physical collision model consisting of several collision spheres with a certain proliferation rule. To be more specific, we only need to apply three simple rules to mimic the natural phenomenon:
Picture1 Brain Coral source link
Picture2 Brain Coral source link
And all the force vectors used in the above rules obey Hooke’s Law, which means their magnitude is proportional to the difference between their distance and the sum of radiuses (F = k*(Distance - RadiuesSum)). So those points can be regarded as a group of bouncing balls connected by springs. Then after implementing these rules to points iteratively, a typical differential growth pattern will be generated. And this generative algorithm is very robust and it ensures the pattern to fill in any 2D boundarys:
This pattern can also wrap complex 3D surface perfectly and the algorithm can be extended to simulate mesh growth:
Here are some laser cuttings and 3D printing results of the geometry generated by this algorithm as a conclusion of the algorithm exploration stage:
Stage 2. Morphogenesis Mechanism Research
After delving into the morphogenesis aspect, we got a deeper understanding of the underlying mechanism of the formation of natural leaves’ curly shapes. From Changjin Huang et al.’s research, we learned that the boundary area of the leave usually grows faster than the middle area, which can be proved by cutting the leaves into slices and comparing their length just as shown in the diagram, this differential growth behavior is actually the key factor for curly shapes. And furthermore, different plants usually have varied growth speed ratios between the edge and middle area which leads to the different types of twisted or curly shapes of their leaves. Inspired by this mechanism, we tried to do some analog experiments to reproduce the natural curly shape. Specifically, we cut some paper strips with sequential lengths and then tried to stick them together with glue. As a result, we got similar curly-leave shapes successfully as shown below:
Apparently, even though the first analog test reproduced the similar natural curly shape, it's still static and not dynamic. And then the second study model was able to change its shape by manually pushing or pulling those strips, but its dynamic feature is still not enough. So we’re trying to find more possibilities from other materials.
Pic1. Three-dimensional morphologies of long orchid petals source: https://doi.org/10.1073/pnas.1811296115.
Pic2. Dissection of twisting leaves into thin strips for growth strain quantification 
Stage 3. Digital Fabrication Process & Final Interactive Prototype
1.Customized 3D printing on pre-extended elastic fabric
With the understanding of the morphogenesis mechanism, we know that if the edge is longer than the middle area, then the sheet will deform a natural curly shape. To make sure the boundary section is longer than the middle section, we first stretched the elastic fabric and fix it on the printing plate and then printed PLA only on the boundary area to maintain the prolonged length. We leveraged customized G-code to print the unique 2D differential growth pattern generated by our algorithm mentioned above. Here are three reasons to use the pattern as the toolpath. Firstly, since the pattern is based on a physical sphere-collision model, the pattern is distributed evenly so that the internal extension force is also distributed evenly. Secondly, the pattern is an efficient space-filling pattern, so using it as the toolpath can decrease total path length which leads to a shorter printing time. Lastly, the differential growth pattern is continuous and contains no hard corners, which is beneficial for good printing quality. After printing PLA on the fabric, the fabric has different stiffness in different areas, so that when it’s subjected to external force in the middle area, it will deform like a natural leaf.
We still explored other variations:
The whole control system consists of two parts. The first part is an Arduino system that contains a gesture recognition sensor and several 64kg servo motors. The sensor can detect the movement when people weave their hands over it in four directions, up, down, left, and right. After receiving the signals from the sensor, the Arduino board can send commands to the servo motors to do responsive behaviors. And the second part is a cable system that is based on an acrylic framework and is in charge of suspending the fabrics and stretching or loosening the fabrics under the control of servo motors. After finishing the controlling system, we tried building a small prototype (50x50x80cm) and also a larger one (1x1x4m) attached on the ceiling of our studio classroom to test its robustness.
2.Controlling system
To make the prototype more interactive and also to get fancier visual effects, we also built an interactive lighting system based on TouchDesig. In this system, the lighting effect can change its shapes and color according to users’ body gestures. About the general logic of the whole system, firstly we used a Kinect camera to capture the body gestures of users and then passed the data to the Touch Design software in the computer, then after a real-time processed visual effect is generated, lastly we project the video onto our prototype by using a projector. Moreover, the input of the lighting system can be other things such as human facial emotions, weather, and so on.
3.Interactive Lighting System
After combining all the systems above, we achieved our basic interactive prototype, which can automatically change its form and lighting effect on itself according to external stimuli. When people weave their hands in front of the gesture sensor, servo motors will pull the fish string, which will deform elastic fabrics. Since the boundary region of the fabric has already been enforced by 3D printed PLA, the fabrics will form a curly leaf-like shape finally which aligns with the morphogenesis mechanism of Differential Growth. In addition, the interactive lighting system can also change itself responsibly to highlight the curvature difference. Finally, when people weave their hands again, fabrics will recover flat again.
4.Final Interactive Prototype
Stage 4. Public Interactive Installation Design
Based on this basic interactive prototype, we want to leverage its interactive ability to develop a design concept for a public installation over Flatiron Plaza in New York City to activate the vitality of the urban environment. We want to install some anchors on surrounding buildings around the plaza and connect steel strings to those anchors to hang up two giant pieces of fabric over the Flatiron Plaza. Those two pieces of fabric are stiffened on the boundary region by using large-scale 3D printing so that we can use motors to tighten strings to make them form curly shapes over the air.
We also built a 1:100 scale physical model to predict its visual effect and influence on the neighboring urban environment, and this also concludes this whole project…
