Modular soft robots_4

PART 4 of 4

For PART 1 click here
For PART 2 click here
For PART 3 click here

For the fourth iteration I built on the 3rd one but this time I inverted the shape of the inflation chamber. Made the walls thicker and and the inflation chambers much thinner and acute. Below is an image of the cross section of this actuator.

trial 5.PNG
This illustration shows the different approaches in trial 3 and 4
trial 4.PNG
cross section view of one segment of the mold. 

I only decided to print one segment of the mold to make de-molding easy.


Unfortunately the Silicone hadn’t cured even after 7hrs!

By the time I got to casting the silicone, the 00-50 grade which we had in class had gotten over. I bought the 00-30 variant since it had a cure time of 4hrs, only 1 more hour than the 00-50 grade. But even after 7+hrs or curing time it was still runny on the bottom end of the cast.

I have a feeling it probably has to do with the cooler temperature on the ITP floor. I think it cures faster at higher temperatures. I plan to recast this actuator on a slightly warmer day and continue this process through the summer break.

Modular soft robots_3

PART 3 of 3

For PART 1 click here
For PART 2 click here
For PART 4 click here

For the third iteration I wanted to model something after the accordion structure.
This video below shows the accordion shaped actuator extending linearly

I took inspiration from this but decided to add some solid chunks of silicone in between to add some stability to the actuation.

trial 3.PNG

As seen in the CAD model above, the cuboidal blocks of silicone do not inflate. They work as internal bracing elements. The blocks also serve as channels for the proposed external restraints to pass through.

trial 3 a.PNG
Mold design for the actuator with the outer part and inner core.
3D printed piece of the mold. The red wire seen on the left is to keep the cantilevered end of the inner core from touching the outer mold part.
I casted only one half of the actuator first and decided to color the other half.
Fully casted actuator, there were a few tears toward the end of the casted piece so I had to cut it in half. (6)

As you can see, this iteration achieves a good amount of linear extension but it inflates in a  rather ugly manner. Also it doesn’t inflate evenly. I later found out from Kari that silicone has a tendency to do that, maybe the wall thicknesses were slightly uneven and once it inflates beyond a threshold, it creates microscopic tears which irreversible alters the behavior of the actuator. (1).gif

I tried inflating the actuator with external restraints (white wire attached to the actuator). As you can see it did successfully turn linear motion into bending motion. I just needed to refine the inflation aesthetics in the following iterations.

Modular soft robots_2

PART 2 of 4

For PART 1 click here
For PART 3 click here
For PART 4 click here

For the second iteration I wanted to test out a helical spring shape for the inner core of the actuator. The idea was to inflate a spring shaped hollow chamber to achieve linear extension, much like stretching a helical compression spring.

I quickly made a spring with thick aluminium wire coiled around a chisel
I put together a mold with these basic components, all picked up from the junk shop. The spring formed the inner core, the aluminium tube is the outer mold and the cardboard piece formed a conical end of the actuator for air inlet.
Actuator is casted and set aside for curing
And now the hardest part was demolding the inner spring. (5).gif
Again, with this one too, I did achieve linear actuation but the whole form was inflating much more than I liked. I was still happy with the overall form of this piece. I’d like to explore this aesthetically in the future. (7).gif

For PART 1 click here
For PART 3 click here
For PART 4 click here

Modular soft robots_1

PART 1 of 4

For PART 2 click here
For PART 3 click here
For PART 4 click here


For my finals for soft robotics class I want to make a system of modular soft actuators which can be combined together to perform different functions.

The soft modular system is an effort to create a plug and play system with soft robotics which is a very fabrication intensive process. I am also thinking of this system as working like an educational tool or toy to introduce kids and adults to the field of soft robotics. This will be to soft robotics, what K’nex is to mechanics or Little Bits is to electronics. Ideally, these blocks should also be compliant with other forms of robotics where they can be easily integrated.

Examples of similar systems

Modular soft robots by EPFL

The modular body

Modular robotic cubes



I had two approaches ahead of me

  1. Multiple modules for a common function
    This approach required me to fabricate multiple modules which would come together to carry out a single function. Like putting together the pieces of a puzzle.
  2. Same module, Multiple functions
    In this case I will make multiples of the same module which can perform differently with simple modifications or combinations. I found this approach to be more interesting and feasible, given the time frame of 3 weeks


The image below demonstrates my idea of a kit with different modular components.

The kit mainly comprises of linear silicone actuators, internal restraints, wheels, end caps, etc.


The different components can be fit together to create different robots or mechanisms as shown in the animation below.


clockwise from top left : Rope climbing robot, gripper, CG wheel, crawling robot.


The chart below explains the motivations and rationale behind the project.

Untitled presentation (2)


Most common actuation methods used in soft robotics are linear (forward and backward) and bending. I realized I need only design a linear actuator and use a mix of external and internal constraints to achieve different kinds of motions. So the actuator module remains same but the variety of motions is brought about by modular restraints.

Update : April 3rd

To test out the idea I designed a mold in Rhino to cast a basic actuator. I also wanted to test the coupler design which would enable the modularity and interfacing between different modules.

mould design.PNG
Dark grey part is the actuator and the light grey covering is a cross section of the mold design
mould design 2.PNG
Final molding assembly file ready for 3D printing



Shown in red in the image above is the coupler in hard plastic which is assembled with the actuator. The coupler has 4 arms which serve as hooks for different restraints. The coupler is slightly oversized to achieve a tight fit with the silicone actuator. In the final design the coupler should be in-molded with the silicone part. The 4 holes seen on the silicone part at the corners is the channel for the internal restraint to pass through.

Update : April 7th



Main body of the cast is ready for molding. I inserted two sticks in each half to form the channel for the restraint.



I also made small holes to let the air out and get the silicone moving more freely.


One half of the mold assembly. The green stick in the center forms the central inner air cavity.



Unfortunately the silicone started leaking from the bottom due to increasing pressure. I decided to let it sit and re-cast it the next day.

Update : April 8th


Was happy to see at least half of the actuator was casted. If anything it was a good run for testing the quality of the cast. Also gave me a much better idea for dis-assembly.


After de-molding, I reassembled the mold and plugged the holes and the edges with hot glue this time.



Parallely I cleaned up the half casted actuator for early testing of the actuation.


Tested assembling the coupler for fit.


I decided to plug one end of the actuator with more silicone and set it aside for casting.


3 hours later…

The actuator is sealed from one end

On the side, I injected more silicone into the mold which was still curing, to fill it up completely and compensate for some of the silicone which leaked at the bottom.


Update : April 9th


Final casted actuator. The red cable demonstrates the restraint mechanism. (6).gif
I was able to achieve some linear actuation but not without overall inflation.



For PART 2 click here
For PART 3 click here
For PART 4 click here

Material Connexion visit

We visited the Materials Connexion library to look at emerging materials we could work with for our finals for soft robotics class.

I primarily looked at Silicone materials as I plan to use Ecoflex extensively for my final project. Listing down a few materials and processes that I am interested in exploring. Some of the materials listed below were found on the online library.

3D printed Silicone
Manufacturer : ACEO


I can use this process for creating intricate channels inside my parts which would otherwise be difficult with lost wax casting or similar process.

Thermochromic Silicone
Manufacturer : Cainir Garment Accessories Co.,LTD


I can see a possible use for this for smartskin development and chromatophoric biomimetic robots.

Thin film solar panel
Manufacturer : Power Film Inc.


I can use this in autonomously powered robotic modules much like soft BEAM bots




Next Nature


cover 2.jpg

I am interested in exploring BEAM robotics for my solar assignment for Energy class. BEAM, to me is fascinating in its biomimeticness and in its handicraft nature of free-form circuit sculpting. I am on a biomimetic / fake nature trip this entire semester. These themes are recurring in two other classes I am taking on soft robotics and micro biology exploration. I am eager to see how all of these could come together down the road.

So for this particular assignment I envisioned a BEAM based underwater robot. This is one category I found to be quite under explored in the BEAM circuit (:p). I looked through several BEAM maker forums but didn’t find any documentation. I decided to build my own based off a simple solar engine. The engine uses a capacitor to store charge from a PV panel and runs it through a voltage comparator and transistor to dump the charge into a single motor once the voltage from the PV rises to a particular potential.

Bread boarded solar engine

This is the engine that makes the robot work. It takes in power from the PV panel on one end and a motor is attached to where the alligator clips are. I thought, once I get the motor to move, I can use it to actuate any kind of mechanism I design beyond that point. I was envisioning something to the effect of a deep sea organism. Very delicateelegant and mystical!

festo - Imgur.gif
FESTO’s smart inversion drive

Pictured above is FESTO’s smart inversion drive system, where a geometric floating object (inspired by origami toy) is filled with Helium and propels itself by inversion.
This, combined with a jellyfish was kind of the mental mood-board I had created.

With these references in mind I set out to ideate on the robot’s form, structure and mechanics.



These are some of the early explorations. I eventually hit upon the idea of using a spring shaft as my main drive system. This is commonly found in the Dremel tool kit and as flexible extensions for power drills or screw drivers.

At the center of that DNA molecule like structure is the spring drive. On top is the motor which connects the drive.

Hardware and parts

  1.  15″ Spring shaft, hacked off a ‘flexible gripper’
  2. 6VDC gear motor
  3. 2mm metal shaft to extend the motor’s axle
  4. 4V circular solar panel

Everything else was found on the ITP junk shelf.

First thing I had to do was test the spring drive working principle. I cropped a segment of the spring off the gripper, taped it with a pin at the center and tried rotating it manually. It worked! and it was enough to create some conviction about the idea. I just had to figure out how to drive it with a motor.

Next step was to extend the motor’s axle. I wanted a dual shaft so I could attach both ends of the spring drive to the same axle. In this case I had to reprint a spur gear from the motors gear box for it to hold the new extended axle.

The top most gear to the left was to be edited and re printed

I remodeled the gear in CAD and got it printed. (1).gif (3).gif
Testing the motor with the extended axle

Next, I glued the spring drive to both ends of the motor. Instead of rotating or inverting, the spring drive was spinning with the motor’s axle, making full circles around the motor. I realized I had to constrain it’s motion in vertical direction to keep it from revolving around the motor and instead make it rotate around itself.
It worked just as expected when I rested it against the table’s edge (1).gif
I braced the middle of the spring with the motors body. This was just what I was looking for but of course had to compromise with the brace running along the middle!


I made the necessary modification to the CAD file (pictured above). I also designed a basic casing for the motor and the solar engine and put it for 3D printing. In the meantime I started working on making a jig to glue the linkages to the spring drive. The linkages were to be glued at 90 degree angles alternately. (2).gif
Laser cutting the linkages


Linkages arranged on the jig, ready for gluing.

Parallely, I started working on sculpting the solar engine. This is the best part of the process for me. Free form sculpting of electrical components is an art. You have to carefully plan out the entire architecture and carefully bend and route the wires for soldering. Doing all this with just two hands makes it more challenging.

Finally! after 4 failed attempts
Engine test (with battery)  

It was time to put it all together. After 4 sticks of hot glue it finally came to life.

I could manage to record it just before the motor axle slipped on the 3D printed gear. For this reason I dint continue with attaching the flaps. This is also pretty far away from actually being water ready. I plan to keep refining this idea as and when I get time. The idea for this kind of radial drive is proven mechanically. Now its a matter of making it waterproof and perhaps scaling it up in power to see if it actually propels itself underwater!





BioDesign worksop

Chris Woebken held a Bio-Design workshop in class. It was a speculative design exercise aiming to visualize bio futures through tangible media. The idea was to gather simple products or materials from a dollar shop and re-purpose them to create and communicate new meaning and ideas. Everyone brought in at least 10 items each.


Full spread of random dollar store items brought in by the classmates. Chris brought along some vacuum formed package shells and different stickers. There were also some random magazine cut outs of a range of news articles.





We were to imagine a hypothetical topic related to our research paper. My selected topic is Gut-microbiome and its effect on the brain. 

So I imagined a future scenario where our world is getting increasingly sterile. We live in air quality controlled clean rooms. So are our offices and cars and movie theaters, everything is perfectly clean. Fruits and vegetables do not grow in soil anymore. Most of our food is grown in labs from cell cultures. There is very little of real, organic matter around. And the lack of helpful bacteria has drained our vitality and spirit.
We survive on constant entertainment and media consumption rather than engaging with the real world and seeking pleasure and fulfillment. Celebrity worship is rampant in a society like this which feeds on entertainment media.


In order to prolong their celebrity stature and physical and mental vitality, only celebrities have access to bio-engineered, probiotics and healthy bacteria. this has given rise to microbial piracy! Microbiome pirates are paparazzi of the future world, who are not only on the lookout for candid  photographs of famous celebrities but also their gut bacteria! This celebrity gut bacteria is sold on the underground market like drugs.


This is quick prototype to sketch out this future scenario by imagining a product, in this case, Jenn Aniston’s gut bacteria sold in form of balls and labeled as artisanal mood enhancer! $66


This is the device used to extract the microbiome!


Silicone casting exercise

For my final project for soft robotics class I am interested in making a soft robot with an exoskeleton.  The idea is that the exoskeleton will protect the soft materials from punctures and cuts. The robot would still be based on soft actuation mechanisms. To test out the idea I thought of making a small 3 prong gripper with an exoskeleton. The plan was to 3D print the grippers exoskeleton in a flexible material and fuse that with the inner silicone gripper in the silicone casting process.


The above picture shows a 3D model of the 3 pronged gripper. The red part being the tough but flexible exoskeleton and the inner light grey part being the silicone casting, this is the part that is actuated. The exoskeleton being flexible, conforms to the silicon inflation.

I took these files to 3D print. But it turned out that the 3D printers were not capable of printing in flexible material. So as a workaround I printed the upper cylindrical part and the 3 flaps separately and thought of joining them with black electrical tape.


Pictured above is a CAD model of the mould for casting silicone, the sloping feature to the right is the spout from where the silicone will be poured.


the silicone casting mould is attached to one of the 3 flaps on the exoskeleton as pictured above and the silicone is poured from top (indicated by the black arrow)


3D printed parts ready for casting!


I poured in the silicone and set the part aside for curing. I only applied mould release to the inner part as I wanted the silicone to adhere to the gripper flaps. I have to admit, the silicone pouring process was extremely difficult as the size of my nozzle was very small. The silicone wasn’t flowing freely into the cast, instead it was just sitting at the nozzle opening after a point which made me think that the silicone was fully poured in.


As you can see, it was a massive failure! But it left me with some valuable learning.

  1. Silicone wont naturally bind to any surface. I should create a physical/mechanical bond to secure the two together. One possible way of doing that is to create holes on the one surface (usually the harder surface) and let the silicone flow through it over to the other side. On the other side the silicone needs to be fixed in place with a slight flange. 
  2. The nozzle for pouring silicone should have a good enough opening, at least 10mm will make it easier to pour in.
  3. Incorporate small runners throughout the mould on the opposite ends of the nozzle to let the air out as you pour the silicone. This minimizes bubbles from forming in the silicone and at the same time helps the silicone flow into the mould. Sometimes if the nozzle is too small to clog the silicone and the mould doesnt have any means of releasing the air the silione will not pour in.
  4. An alternative way of casting this would be to first pour the silicone in its mould and then secure it to the flap, that way i could have easily poured all the silicone into it’s mould.

Big Box store visit

Photo from Arnav.jpg

Amena and I visited Target to do some research on some soft goods for inspiration. The intention of this exercise was to identify products that use soft materials to use as idea triggers for your final class project.

Since I am interested in making a soft robot with an exoskeleton to protect it from wear and tear and getting punctured or cut, I was on the lookout for similar products with a mix of hard and soft.

To be honest, I didn’t find too much variety. Most of the goods were essentially soft toys. Below I am listing down few of the things that were noticed. I might also add to this list, references found on the Internet or other sources.

 1. Soft + hard shell sky projector toy
A soft elephant with a hard shell on top. Underneath the hard shell is high power LED which projects the star shaped cut outs on the shell all over the room. This inspires me to think of my retracting robot from an assembly point of view. To first cast the silicone, pack all the electronics inside it and then glue the exoskeletal parts at the end.
2. Soft toy puppy with embroidered graphics
IMG_20180304_195438959Not much going on here. But the graphics treatment had me thinking about incorporating surface graphics and and textures in my robots. I am eager to explore combining different castings with varying colors and textures on the same robot to differentiate its various parts. I will need to first cast all the colored/textured components separately and then find a way to insert mould them into the larger and final robot body. This will add some depth to the otherwise bland looking soft robots that we normally see.
3. Suction cup balls
img_20180304_200127739.jpgThe classic childhood toy. The suction mechanism is very interesting when I think about it in a robotics context. Would be interesting to explore mechanisms where the suction can be introduced and taken away, in a programmed way. I think FESTO already does that in their octopus tentacle inspired gripper.
4. Modular toy
img_20180304_195835603.jpgA modular slithery toy! That actually gives me a very exciting idea of building modular blocks of soft robotic actuators to build custom robots! Much like Oscar but without the creepiness and gross living tissues, using silicone and other inflatable materials instead. I might decide to take this as my final topic.
5. Nice texture
img_20180304_200313696.jpgI like the textural quality of this product. I can imagine a wall mural with dynamic, kinetic  hair like structures, each actuated individually and can be programmed to create soothing wave like patterns!
Few other products noticed..

This slideshow requires JavaScript.




For our Kinetic Energy assignment, I collaborated with Dan Oved to build something that would turn muscle power into usable electricity.


We started off by brainstorming on a few concepts for generating light from human motion.

Hand cranked music box

The first thing we explored was turning a hand-crank music box into one that emits light in sync with the music.

We would connect a motor to the end of the crank shaft and have that generate electricity.

Wheel chair powered light

We explored using the wheelchair on the ITP floor for kinetic energy – it could work in one of two ways.

When the user pushes on the wheel, it would rotate a DC gear motor attached via a belt, to generate electricity.

Or, when the user pulls the break pedal, it would move a gear motor attached to a rubber wheel down to have the rubber wheel touching the wheel of the chair, the resistance causing it to break and generating electricity.

Electric Poi

Our final idea came from something I’d seen in India, a variant of slingshot where farmers spin a weighted ball attached to a rope around in circles, in order to keep birds away from their crops. Back home its called a gofan but it closely resembles a Poi performance art prop. We thought of turning this swinging motion into kinetic energy. There would be a handle with a rope and wire coming out of it connected to a ball with a light in it. The rope would be physically attached to the shaft of a gear motor which would be inside of the handle; when the ball would be swung around, it would cause the shaft to rotate, thus generating electricity. This electricity would be outputted back to the wire which would power the light.

Poi performance

We then realized this would likely get tangled, or have difficulties rotating the shaft if it was attached directly to it; instead we decided it would be better to attach the rope to a plate or L bracket attached to the shaft, which would increase the torque and force the shaft to rotate.

We decided to go ahead and test this idea for feasibility. We did some calculations to narrow down on the exact motor we would require in terms of RPM and torque. The main components of the math were the length of the string and the weight of the light object attached to it. As soon as the motor arrived, we attached a weighted object to it with a coupler and a wire. We realized the wire wasn’t actually moving the motor shaft but rather just spinning at the very point where it was tied to the motor. It snapped in no more than 4-5 rotations.

We had to think of something else! Thats when I recalled of seeing a make shift and re-purposed, rural Indian toy where these kids in rural areas would essentially find anything which is round and attach it to a stick and run around the village rolling that wheel on the ground. We thought of building on that with a motor at the center of that wheel which powers an LED strip stuck to the periphery of the wheel itself. This meant that we don’t have to worry about mounting a slip ring in between the motor and the shaft or light.


Selecting the motor

To power the lights, we would force a DC gear motor to turn and tap into the electricity it generates. We chose a 127 RPM Mini Econ Gear Motor from ServoCity, because it had optimal RPM and a low torque of 9,602 kgf-cm.

The 127 RPM dc gear motor we used from Servo City

We would use a .770” Pattern Clamping Hub which would attach to the shaft and rotate to generate torque:

The clamping hub

CAD design

Servo city had provided detailed 3D files for the motor and the coupler on their website. We thought it would be best to plan, troubleshoot and refine the idea in CAD itself to save on time and fabrication. We used Rhino to design the wheel around the selected hardware and mounting brackets.

Cross section of the wheel at the spokes

Prototyping the circuit

Next it was time to build a circuit that would simulate the power generation. We used 4 diodes as a bridge rectifier so that no matter which way the motor was turned, the current would flow through the circuit in the same direction. We used a bunch of capacitors in parallel to smooth out the current and store energy when the wheel stopped turning, enabling the lights to stay on. We used a 12v white analog LED strip since it has the same voltage rating as the DC Gear Motor.

The prototyped circuit with a bridge rectifier

We were able to test the circuit successfully. Turning the motor in either direction generated power to turn on the lights:


We converted this 3d model into slices of 3/4 inch to cut it on the CNC Router

Then we glued it all together and let it sit overnight

The engine mounted to the wheel. This would allow the circuit to rotate with the wheel without the need for a slip ring.

We found a tripod leg on the ITP junk shelf and thought it would be perfect as a stick that drives this wheel. Plus it was already well finished and accommodated the motor’s mounting bracket perfectly

match made in heaven!

Fabricating the circuit

Next it was time to build the circuit that could withstand a series of rapid rotational turns. We designed something simple in EagleCad

The circuit with the bridge rectifying diodes and 47uF capacitors in parallel


The circuit with the capacitors in parallel and bridge rectifying diodes. Yellow wires would go to the motor and accept either polarity, and red and black wires would be power and ground for the lights.

We tested the circuit with an external 12V power supply, and also tested reversing the polarity of the power supply to see if the bridge rectification worked, and were glad to see that it did.

Sticking the LEDtape and routing the wire. We also used 3M’s friction tape to add some traction

–Putting it all together–

After coating the wheel with black spray paint, it was time to assemble everything and test it all.

The spray painted wheel with circuit attached
Yellow wires pass through the center and are attached to the DC gear motor’s power and ground.
The shaft of the motor coupled to the rod; when rolling the wheel the torque would cause the engine to generate power.
Getting ready to test everything

Dan testing the wheel
It worked!