Polyurethane foams and 3D-printed lattice structures were covered in wax and tested at various temperatures. The results indicated that wax-coated structures have the potential to act as a much more controllable medium for robotics, allowing for controlled changes and repairs to be made to the structure’s stiffness modulus through temperature change.

Cheng, N.G., Gopinath, A., Wang, L., Iagnemma, K. and Hosoi, A.E. (2014), Thermally Tunable, Self-Healing Composites for Soft Robotic Applications. Macromol. Mater. Eng., 299: 1279-1284. https://doi.org/10.1002/mame.201400017

Paper 2: Puffy, a friendly inflatable social robot (S21 R75)

This paper (and supplementary video) discusses the purpose and features of a Baymax-esque soft robot designed to improve children’s learning environments. The video demonstrates a classroom application of the technology using a conceptual prototype, however neither it nor the article go into technical depth. That being said, it is made clear that the multi-sense interactivity and target aura of Puffy is largely possible due to it being a soft robot.

Alessandro Ubaldi, Mirko Gelsomini, Marzia Degiorgi, Giulia Leonardi, Simone Penati, Noëlie Ramuzat, Jacopo Silvestri, and Franca Garzotto. Puffy, a friendly inflatable social robot. In Extended Abstracts of the 2018 CHI Conference on Human Factors in Computing Systems, CHI EA ’18, 1. Association for Computing Machinery, 2018. doi:10.1145/3170427.3186595.

Paper 3: Electroactive textile actuators for wearable and soft robots (IEEE)

Dielectric elastomer actuators (DEAs) are thin, multilayered membranes that physically deform as a result of a voltage being applied. This paper tested the relative change in top surface area as a result of small voltages (9kV and 6kV) being applied to multiple DEA configurations. This technology can harness this deformation to create specific movements and ultimately achieve actuation from a voltage supply.

J. Guo, C. Xiang, T. Helps, M. Taghavi, and J. Rossiter. Electroactive textile actuators for wearable and soft robots. In 2018 IEEE International Conference on Soft Robotics (RoboSoft), 339–343. April 2018. doi:10.1109/ROBOSOFT.2018.8404942.

The objective of this project was to create an ecosystem of swimming jellyfish robots that could actuate soft tentacles underwater to produce movement. In terms of an artistic vision, we wanted to create a peaceful and calming piece. We aimed to create two types of jellyfish: one jellyfish that is fully actuated with 3 independently moving legs, and one jellyfish that has a single actuator and moves by catching the current. By showing these two jellyfish together, we can draw contrast to their differing modes of movement; one jellyfish being self-sufficient and the other depending on the water’s movement.

The fully actuated jellyfish represents an interpretation of bio-mimetic robots where the form of the robot closely resembles the structure of natural organisms, but the functionality is much more complex. The goal of the single actuator jellyfish was to produce a soft robot that mimics both the form and function of jellyfish in nature, which primarily ride ocean currents for locomotion with minimal intervention through movement.

Creative Design Opportunities

We used molded soft silicone to make the jellyfish. The soft material allowed for each leg to be flexible. The soft material allowed channels to expand when filled with water, which caused a leg to bend. The soft material also meant that the jellyfishes would be gently moved and rippled by the water currents, and give the effect of floating.

Outcomes

We successfully designed and fabricated two jellyfish.

The biggest success was the pronounced movement of the jellyfish legs of the 3-legged jellyfish. The motor pump combined with the small water channels allowed for a very significant bend in the legs. Furthermore, we were able to refine our molding and bonding process to produce a final jellyfish without air leaks or accidentally sealed channels. We adopted design elements and techniques from the JenniFish paper cited below. We created a two part mold for the top piece of the jellyfish, which was bonded to a thin bottom flap. We used clamps to ensure a tight fit for the two part mold which prevented leaks.

Areas of improvement include the buoyancy and displacement of the jellyfish. The silicone molded part is too heavy to float once the inside is filled with water. We added a balloon to compensate for the buoyancy so that the jellyfish would be more neutral with the water and float. However, it was difficult to get the inflation of the balloon perfectly so that the jellyfish would float suspended in the water instead of either sink down or bob to the top of the water. Furthermore, we had hoped motions like pulsing would create more vertical motion such that successive pulses would allow the jellyfish to swim up with a significant displacement from the starting position. However, the jellyfish mostly stayed in place. For the 2-legged motion, we did see noticeable horizontal displacement and the jellyfish was propelled in the direction of the non-actuated leg. Thus, we assume that the problem with vertical displacement is related to the buoyancy issue described earlier.

In addition, the single-actuator “passive” jellyfish could have had a more distinct movement. There were a few air leaks in the legs, and the flap was slightly too thick, making it difficult for the thin legs to bend it.

Video

Citations

Frame, Jenny. “Self-contained soft robotic jellyfish with water-filled bending actuators and positional feedback control.” (2016). https://fau.digital.flvc.org/islandora/object/fau%3A33675/datastream/OBJ/view/Self-Contained_Soft_Robotic_Jellyfish_with_Water-Filled_Bending_Actuators_and_Positional_Feedback_Control.pdf

Frame, Jennifer & Lopez, Nick & Curet, Oscar & Engeberg, Erik. (2018). Thrust force characterization of free-swimming soft robotic jellyfish. Bioinspiration & Biomimetics. 13. 064001. 10.1088/1748-3190/aadcb3.

Chai, Katie. Soft Robotic Jellyfish, Instructables. https://www.instructables.com/Soft-Robotic-Jellyfish/

Technical Documentation

Fully Actuated Jellyfish

Passive Jellyfish

CAD

Drive Link: https://drive.google.com/drive/folders/1uOuozLeJPmjkVByYSKEhGGOgxZAxN89-?usp=sharing

Contents

Hand Pump Test: model of a single actuated tentacle with a hand pump

Tension Wire Test Arm: Wire driven, manually actuated tentacle

Tension Wire 2: Second iteration of tension wired tentacle

Jasmine Jellyfish: First design for full jellyfish with 3 tentacles and one pump

Jellyfish 2.0: Second and final design of 3 tentacle jellyfish with independent tentacle actuation via 3 pumps

Smol jellyfish: Passive, single pump large flap jellyfish

Source Code

https://drive.google.com/file/d/13dDZ4bJURZZgL-e9u6N5yBAhruQPjPO6/view?usp=sharing

Contributions

Both partners worked on both jellyfish design and fabrication fairly equally. Jasmine wrote the code.

VoidCrawler – Jason

The main objective of this project was to create an alien creature that embodied the theme of “Void” through peculiar movement and anatomy. I wanted to offer people an experience of encountering a foreign, alien organism that blurred the lines between lifelike and lifeless. The project utilized silicone to emulate fleshy appendages and motion, while making use of rigid material to create a protective shell.

Rnav – Inbar

My objective was to recreate Whitesides’ soft robot successfully, in order to then create another version of this robot that is scaled up to the dimensions of my pet rabbit. The purpose of the scaled-up version of the robot is to be feathered in one scene of my documentary project, which premise is centered around my hopeless attempts to restore the lost libido of a castrated pet. The design of my soft robot has only one variation from the original one – another round chamber that will be filled with a smell-induing liquid, for the purpose of drawing my rabbit to it. Thus, my objective was not to change the original design of the soft robot, but rather to utilize its aesthetics and the connotations of its materials to then showcase it in the context of my video project. The current title of my project is “Rnav”, which is a misuse of English letters that resolve in what sounds like the Hebrew word “Rabbit”.

Creative Design Opportunities

Jason

The use of soft silicone material allowed me to design and fabricate appendages that mimicked the texture of slimy and shiny creatures. Additionally, the silicone’s behavior when actuated created a slow and steady motion that imitated how animals get up from rest.

The versatility of the silicon also allowed modifications to be made during fabrication. Although not an intentional design choice, the use of the fabric within the clear silicone appendages created an interesting visual effect that resembled that of jellyfish. Furthermore, the usage of dyes within the silicone parts allowed for more creative freedom by creating really dark parts that match the theme of Void.

Beyond the silicone molds, the use of rigid parts through 3D printing incorporated some mechanics within the design. I was able to design a central unit where multiple appendages could be attached to create an n-legged creature. Similarly, I was able to use color with this material as well, specifically black, to produce a shiny and hard product that perfectly resembled a hard outer shell of an organism.

Inbar

The design process of Whitesides’ soft robot required me to learn several new skills: silicone rubber casting and bonding, parametric design using SolidWorks, and 3D printing. This is the first time that I’ve experimented with either of these skills, and the intense engagement with them during the course has led me to acquire enough knowledge so that I can continue to use them in my future projects. I find this very gratifying since silicone rubber and mold making is a scope in sculpture that I have always wanted to experiment with but never did because I felt I lacked the needed abilities and/or knowledge.

Silicone is an inspiring material that is fascinating to work with because of its skin-like qualities, and the different signifiers it can evoke in different contexts: body implants, props, sex toys, etc. However, the design and manufacturing process with silicone needs to be very precise and well-planned, especially when trying to mimic an existing robot from a scientific paper that does not include instructions or exact dimensions of each part. Such a slow and precise work process is very different from the creative processes to which I am accustomed in my artistic practice, and I believe this is the reason I had to go through a few failed fabrication attempts until I became accustomed to the orderly work process.

Outcomes

Jason

The final outcomes of the project were a fully assembled Void Crawler, albeit with only two functioning appendages. The sequence developed for the Void Crawler consisted of a basic rising up motion followed by actuation for just one appendage at a time.

The greatest success of this project was the design of the soft silicone appendages and the use of fabric during fabrication. The initial design choice of making the appendage a stretched-out hexagon was arbitrary but later showed interesting behavior as the chamber design was modified. The varying length of the thin chambers from end to end proved to create a desired “joint-like” actuation. Although this was assumed to be a result of the chamber pattern, there was no formal test to validate this claim. Upon additional research, a fabric embedding technique was found. It was implemented and provided a more consistent and circular curvature. Although some papers mentioned specific materials, regular and thin fabric performed really well.

One of the greatest failures within the project was bonding. Although in the end I managed to have two fully operational appendages, the bonding process proved to be too inconsistent to guarantee full bonding every time. There were several attempts to fix this issue. The latest solution with mixed results consisted of changing the shape of the tube port to square to allow slicing of a thin base layer. A full thin layer would be spread over the intricate top mold and then bonded with the thin bottom layer.

The current state of the project is very much a prototype. There are many ways this project can be improved upon. One of the key areas for improvement would be with the bonding. The research papers used in this project mentioned the use of “bonding surfaces” to ensure proper bonding. The current molds could be modified to include a new bond design that would incorporate this idea. Additionally, the lengths of each chamber and distance can be modified to see if there are more interesting movements that can come from the hexagonal shape. An idea mentioned early on in this project was the incorporation of an additional chamber within the center unit to initiate a vital organ, such as a brain. This idea can be further developed in the next iteration of this project.

Inbar

During the design process, Jason encountered a post online that documented a similar studential attempt to recreate Whitesides’ soft robot. During their fabrication process, these students decided to slightly change the robot’s design in order to allow for easier fabrication and bonding (they expanded the width of the inner chambers and also expanded the distance between the chambers). After several design attempts, I’ve decided to follow the same path and tried to recreate their design rather than the original Whitesides one.

My final soft robot bonded successfully except for one small slit in the bonding area, which I detached through a bubble test and resolved using epoxy glue. It was also pretty easy to understand and recreate the choreography that is needed to create crawling locomotion using the movement code – I followed the rhythm from a documentation video the previously mentioned students uploaded to that post. All five parts of my robot inflated successfully and the robot seemed to crawl forward through two consecutive loops of movement – until I pressed one of the legs and it exploded (as can be seen in my documentation video). Two more legs ended up exploding later in the showcase event, which perhaps means we used too-high air pressure. However, I do perceive this as a positive result, since the documentation video indicates that during the two movement loops that the robot completed before the first leg exploded, it did manage to move forward.

In the next few months, I plan to attempt to create a scaled-up version of this robot, and then use it while filming the scene I’m planning for my documentary project. I’m planning to scale up the same final design and stick to the same silicone casting protocol but will have to create the molds differently, due to the size limit of 3D printers. As a first step, I will try to fabricate one large-scale mold using wood and laser cutting. Another future step would be to identify and achieve the correct liquid to use within the added camber I added to the design, which might need to be the urine of a different rabbit or other substance that contains the appropriate pheromones.

Documentation video

Jason

Inbar

Citations

Jason

The project was almost entirely original, mainly drawing on ideas regarding fabrication and application found within many academic papers on the topic of soft robotics. One influential paper that helped in understanding the design and fabrication process, however, was the Whitesides paper on an untethered soft robot design. This paper provided more context for fabrication problems and solutions. Additionally, the paper on this starfish robot provided the inspiration for a central unit to attach all appendages.

Tolley, Michael T., Robert F. Shepherd, Bobak Mosadegh, Kevin C. Galloway, Michael Wehner, Michael Karpelson, Robert J. Wood, and George M. Whitesides. “A Resilient, Untethered Soft Robot.” Soft Robotics 1, no. 3 (September 2014): 213–23. https://doi.org/10.1089/soro.2014.0008.

Zou, JiaKang, MengKe Yang, and GuoQing Jin. “A Five-Way Directional Soft Valve with a Case Study: A Starfish like Soft Robot.” In 2020 5th International Conference on Automation, Control and Robotics Engineering (CACRE), 130–34, 2020. https://doi.org/10.1109/CACRE50138.2020.9230177.

Inbar

Hao, Bohan, “Soft-robot Design and Fabrication,” cargo collective, May 2013, https://cargocollective.com/bohanh/Soft-robot-Design-and-Fabrication.

Shepherd, Robert F., et al. “Multigait soft robot.” Proceedings of the national academy of sciences 108.51, 2011, https://doi.org/10.1073/pnas.1116564108‏

Technical Documentation

Inbar – CAD

Group Member Contributions

Jason:

• Updating code to work with the project specifications
• Testing thick chamber designs
• Bonding/Slicing technique

Inbar:

• Locomotion
• Testing thin chamber design

Our objective was to replicate microfluidics experiments from the paper ‘Venous Materials’ at a larger, more wearable scale, by using molds, silicone, and traditional bonding techniques. We wanted to explore various aesthetic effects of moving liquid from a silicone repository through a set of channels, where the motion would be actuated by physically pressing on the channel. We also wanted to explore reversible color change, caused by different colors of liquid traveling through separate but overlapping channels. Additionally, we wanted to explore how we could design our silicone microfluidics piece such that liquid could get locked into a specific area, even when pressure was released from the repository.

Creative Design Opportunities

When we first began this project, we had initially planned to create a silicone piece for athletes to wear that uses microfluidics to visually indicate pressure applied to different parts of the body. For example, a headgear could have a liquid repository that, when pressed (after the side of the head was banged), would release liquid through a channel into a “locked state”. We wanted this effect to be reversible; we were aiming to design our silicone piece so that applying pressure at various points could cause the liquid to retreat back from its locked position into its original repository.

Above all, we were aiming for a strong visual component to our silicone piece. We wanted our liquid to flow through channels in an intricate, beautiful way, and incorporate effects such as color change. In short, we wanted our piece to not only be functional, but also aesthetically pleasing and invoke a ‘surprise effect’ when the liquid traveled through the channels upon impact.

We also realized that because we wanted this piece to be wearable, the individual layers making up the full piece needed to be as thin as possible, so that when we stacked them, the overall piece would be flexible and allow free motion of the body.

Outcomes

At first, we decided to use Ecoflex 45 for our experiments. We started by testing simple motion of liquid through a short channel, which was successful. We did this with two layers: one which was a piece of silicone that contained deformations for the repositories and channel with a floor underneath, and the other was a piece of silicone that served as the ‘ceiling’ to this first piece. We bonded the two pieces and injected liquid from the top of each repository using a .5 mm needle.

We then moved on to testing an overlap effect by essentially creating two of the first piece we tested, and bonding them on top of each other. We injected blue liquid into one side and red liquid into the other, with the blue liquid accidentally being much lighter than the red liquid, seeing as we didn’t realize how volume of colored liquid impacted how dark the liquid was. Thus, we could see that on the side with the red liquid, the liquid going through the channel appeared bright red, however on the side with the blue liquid, we saw that the liquid appeared purple because of the red directly underneath. In the video, we can see the actual blue color in the square repository, where the red isn’t overlapping. Thus, our overlap effect was successful.

Afterwards, we tried to scale up our design into a full wearable piece with simple actuation: press the liquid repository in the palm and liquid would travel up to the fingertip. However, when scaling up, we realized that when we were making the liquid repository and channel bigger, we also had to adjust the thickness of the silicone accordingly, otherwise pressure wouldn’t be concentrated in such a way as to make the liquid move.

Next, we experimented with a locking mechanism. We wanted to imitate the mechanism from ‘Venous Materials’ here:

After viewing this effect several times, and prototyping with paper, we decided we needed four layers to imitate this effect. The first layer would be the standard “repo + channel” layer, with deformations for both and a floor underneath. The second layer would be a .5mm sheet of silicone with 3 vertical grid line punctures that were 15mm long and 2mm thick, the third layer would be a .5mm sheet of silicone with 3 horizontal grid line punctures that were 15mm long and 2mm thick, and the fourth would be the ‘ceiling’ layer. Liquid would travel from the repository through the channel in the base layer, then flow through the vertical punctures in the second layer and travel along the ‘floor’ of the third layer surrounded by the walls of the horizontal punctures. Surface tension, we hoped, would keep some of the liquid trapped in the third layer, with the rest of the liquid flowing from the third to the first layer. After discovering that 2mm thick grid lines were too thick to enable much surface for the liquid to bond to, we settled on .5mm thick grid lines.

Since this required some very fine bonding and detail work with the molds, we initially experimented with laser cut .5mm Yupo paper layers for the grid layers. This ended up working poorly because the Yupo paper bonded very badly to the silicone top and bottom layers, which meant liquid would consistently leak out of the ends. Additionally, it was a bad idea to inject into the repository from the top because upon pressing the repository, liquid would gush out through the injection hole.

Although we ended up getting an effect similar to what we wanted, as shown below, it was dubious as to whether this was because the Yupo paper was getting stained, or the liquid was actually getting trapped. Thus, we moved on to actually using silicone for all 4 layers.

When we used silicone for all 4 layers, we ran into a lot of issues related to bonding, where liquid would easily leak from the channel/grid lines out of the sides. Our first semi-successful attempt involved the use of DragonSkin 20, hence why we decided to make the switch to DragonSkin after using EcoFlex 45 for all of our previous attempts. This we considered successful because as intended, liquid was traveling orthogonally to the initial direction of the base channel, and then it was getting trapped in the third layer due to surface tension (note the dark blue stain among the lighter blue). However, the effect wasn’t that pronounced, so it wasn’t a complete success.

From this, however, we learned that we could devise a mechanism where liquid would travel from a channel in the base layer through a puncture in the layer above it into a second channel in that same layer (include an image to illustrate what you’re talking about). This meant that we could potentially achieve the color overlap effect without the traditional approach of channel + ceiling + channel + ceiling, instead doing channel + channel + ceiling, which meant that our overall silicone piece could be much thinner.

When we put this into practice, however, our results were poor because it proved difficult to bond the layers together. There was the pitfall of using too much silicone to bond, and having the second layer channel be sealed up with silicone, or using too little silicone to bond, and having liquid from the base layer mix with the liquid that was in the second layer, because there wasn’t enough silicone to bond the part of the sheet that separated the base layer and second layer, or having the liquid leak out from the sides where the puncture was located.

Our final experiment involved a small tweak to the first approach where we stacked channels on top of each other. Instead of doing channel + ceiling + channel + ceiling, where each “channel” piece would be face-up with the floor on the bottom, we decided to do ceiling + channel A (face down) + channel B (face down), where the floor of channel A would serve as the “ceiling” of channel B. In order to reduce the likelihood of liquid colors mixing, which was still possible if the liquid repositories were stacked directly on top of each other and a needle accidentally punctured the floor of the topmost repository, we also decided to design the layers in a “wristwatch” style. This meant that one repository was located on the left side of the wrist, and the other repository was located on the right side of the wrist, so if the needle accidentally punctured the floor of either repository, there was no way for color to mix between the two.

This method turned out to be our most successful method, though we still had a lot of issues with bonding. The way we solved this was by using a paintbrush to apply an even layer of silicone. Additionally, the Dragonskin 20 turned out to be too opaque to clearly see the different colors of liquid when layers were stacked, so we reverted back to using Ecoflex 45. We discovered our issues with Ecoflex 45 in the past had been due to poor bonding and poor injection.

Here are our most successful experiments

Key Takeaways

For injection

• Before injecting, turn the needle upwards, tap on the side of the liquid chamber a few times, and slowly push up to release any air. This won’t release all of the air, but it will release most of it, and keep the repository from ballooning up when injected with the needle
• Inject from the back side of the repository and slowly shimmy the needle into the actual repository before releasing liquid. Additionally, don’t inject too close to the repository’s ceiling, but also don’t inject too close to the floor. Otherwise, it’s easier for liquid to leak out after injection

For molding

• When making molds with a certain depth for the channel, make sure that there’s a decent amount of thickness for the floor. Ie if your mold has 2mm deep channels, the floor underneath those channels should be at least 5mm, meaning the entire mold should be 7mm tall. Otherwise, the printer won’t register the floor as existing.
• Use the vacuum degasser to remove air bubbles

For channel/repo design

• Channel volume (area * depth) should be less than or equal to the volume of the repository
• If you’re scaling up the area of your repository (i.e. by 1.5x), you need to scale up the depth of the repository (i.e. by the same 1.5x)

For bonding

• Use a paintbrush to get an even layer
• Potentially create a slightly raised silicone “wall” around your channel in the base layer that combines with indents in the ceiling layer, so that when bonding, you don’t need to worry about silicone leaking into the actual channel and can thus be more generous with applying the bonding layers

Overall, a lot of mistakes were made, and a lot of trials were needed. Here are the trials we did.

How the Molds were fabricated

In Fusion:

• We started by modeling what we wanted the silicone piece to look like
• If the silicone piece had divots for the channels, we would flip it 180 degrees so it was upside down
• We then made a larger block that surrounded the piece and had its ceiling level with the ceiling of the silicone piece (or flipped piece), and had a height that was ~5mm additional to the height of the silicone piece
• We then used “Combine”, used the larger block as a Target body and the silicone piece as the Tool body, and used the cut operation, keeping the tool for volume calculation purposes

In order to have multiple molds combine to create one silicone piece (in the event that the printer can’t handle the entire mold), we would slice the mold at a particular space (preferably where there wasn’t a lot of intricate channel work happening) and then create hinges so the two edges of the ‘broken’ mold can connect into one.

Technical and Artistic Contributions

Sunjana

• Engineered and cadded up molds for grid locking mechanism
• Engineered and cadded up all large-scale molds (puncture overlap, wristwatch, initial single-channel with palm repository)
• Engineered and cadded up small molds for initial pressure and color change tests

• Constructed and bonded all silicone pieces for grid locking tests, initial pressure tests, and color change tests

Jina

• Designed the look of the channels and how they would overlap
• Engineered and cadded up small molds for overlap using final channel design
• Engineered and cadded up molds to test shape change due to pumping air
• Engineered and cadded up molds to test Tesla locking valve

Sunjana and Jina worked together on bonding the molds for the puncture overlap mechanism, the wristwatch overlap mechanism, and the smaller version of the wristwatch overlap mechanism (channel + channel + ceiling). Sunjana primarily worked on molding and injecting while Jina worked on bonding. They also worked closely together on ideation.

Final Documentation of Prototypes

When we first started working on our project, we tested simple mechanisms through different techniques like using paper to even just having an extra layer in between. In regards to the paper, the material would stain and not be affective.

https://youtube.com/shorts/uTZhcKelyHE?feature=share

https://youtube.com/shorts/xj96ZMnt1KQ?feature=share

In the video above, you can see the earlier iterations that we created. Due to the video uploading as a short (it can not embed into wordpress). We tried to upload as a normal video but it wouldn’t let us.

In the final prototype, we tried to perfect the bonding as it would affect how the patterns would appear.

https://youtube.com/shorts/YuYRkwfu4b0?feature=share

https://youtube.com/shorts/av7hcFXaE_A?feature=share

Citations

Hila Mor, Tianyu Yu, Ken Nakagaki, Benjamin Harvey Miller, Yichen Jia, and Hiroshi Ishii. 2020. Venous Materials: Towards Interactive Fluidic Mechanisms. In CHI Conference on Human Factors in Computing Systems (CHI ’20), April 25–30, 2020, Honolulu, HI, USA. ACM, New York, NY, USA 15 Pages. https://doi.org/10.1145/3313831.3376129

Molds

Video: https://drive.google.com/file/d/1TEdhAqjEpIGa3L61PbP-pYjLCAwTackj/view?usp=sharing

The next steps are to clean up the wiring, code some more motions, and finalize our piece.

We first looked at whether our patterned designs work! Below is the more rigid design that we created. The reservoir was too small so it did not fill all the way, but with some force you can see the pattern. There are also bubbles with the liquid which we were able to figure out a mechanism to get rid of them.

For this design, we did the math wrong so the layers were extremely thick and didn’t show both of the designs together as much as we had hoped. So for our next 3d prints we fixed that. It was interesting to see the faint blue pattern though which gave our team much hope for this concept.

As we continued to test, we realized that we need to tap and take out all of the air so that there wouldn’t be any bubble in our reservoir.

After testing our smaller tests, we realized that it would be better to change the design slightly to better fit. Below is our final concept/design.

Alternatively, our team was trying out several experiments that involved both liquid repositories in the same plane, where liquid would travel through a channel from one repository, and enter a channel in another layer, so we could have “overlapping channels” with very little obscured because the layers between channels are thinner. However, we discovered this is extremely hard to bond, likely because the second layer was too thin (silicone kept moving into the .25mm channel). It’s going to be simpler to just layer channels on top of each other, and the color can still be seen pretty well.

It’s been a massive inconvenience having been quarantined and barred from campus facilities from the 12th until the 22nd, however we’re hopeful that we can achieve some meaningful result with the molds we have printed.

I took Jina’s design from this morning’s class and turned it into a set of 6 molds, or 3 layers. Below is what the design will look like in silicone form. It is around 6 inches long in total. I had to make tradeoffs between how many channel elements I could include, how much liquid I had available in the repository, and how thick I wanted the layer with the liquid repository to be.

The bottom-most layer contains the two liquid repositories, and is 2.5mm thick. The repositories themselves are 1.5mm deep (I increased the area of the original repositories I was working with by a factor of 1.5). The channels they’re attached to are 1mm deep. This layer also contains the vein-like channels, with ringlets attached to them.

The second layer contains the rectangular channels. A puncture in this layer allows liquid to flow up from the bottom-most layer, into the rectangular channels, which are .25mm deep. This layer is .75mm thick.

The last layer is the ceiling layer. It is placed over the second layer to close off that channel, and is .5mm thick.

Overall, the rectangle is going to be 3.75mm thick, and the amount of material above the two liquid repositories will be 1.25mm.

Below are the 6 molds. Each of them have space for wedges that are 3mm by 4mm by 4mm.

For control, we directly powered the motors with the 5 volt power source. Next week we intend to use the Pico to individually actuate each motor. However, we need another motor controller chip since each chip only has 2 channels.

The latest iteration of the Void Crawler has seen changes in its air channel geometry along with a new slicing location. Although these changes may seem minimal, they seemed to greatly improve both the fabrication process and performance.

From this test, it seems that the flat base helped in ensuring a good bond between both parts compared to the previous chamber-to-chamber contact. Although this part did fail, its failure was contained to a single point rather than a separation of chambers. Once again, embedding the fabric within the base part allowed for more curvature. Its movement is exactly what I am looking for in bringing the Void Crawler to life. With the video, it seems it can easily and securely interact with the centerpiece. I propose for the next test to update the centerpiece to have four ports and test locomotion.

The centerpiece has been updated to include four ports and a locking mechanism. The next steps are to fully assemble the VoidCrawler and practice locomotion with the pneumatic system.

Inbar’s Update

After the previous attempt to create the entire robot, which failed mainly due to difficulty in bonding, I changed the structure and cut plan of the molds so that the tube opening is now on the top face of the robot, and all internal texture is only at the top part so that the bottom part is a thin and smooth layer (which also contain the fabric). That way, the bonding is done by coating the entire bottom part with a thin layer of silicone and then placing and aligning the top part on top.

The bonding in the current version was indeed successful – a bubble test in water showed only one bonding malfunction at the edge of one of the legs – which was easy to solve with epoxy glue.

The airway cavity of one of the distant legs was blocked during bonding, but it was easy to fix by punching the blockage with a needle.

All five parts of the robot show a nice curvature while inflating, as long as I apply pressure to close the opening around the inflating tube. I made an attempt to set the tube in one of the openings with two layers of epoxy glue and it seemed to work – as you can see at min 2:30 of the video.

The next two steps:

1. To program the choreography of the air pressure system in order to produce crawling locomotion.

2. In case I achieve crawling locomotion that is satisfying enough, I would like to plan the fabrication of a scaled-up version, that should be around 70 cm long (27.5 inches), for the purpose of filming the scene with my pet bunny.

Documentation video: