Design

The Fabric-Reinforced Controlled Motion Robot (FRCMR) is a soft robot that mimics the basic locomotive methods of both an earthworm and an inchworm.  It is comprised of three silicone bladders arranged in a triangle. Each bladder is divided into two sections allowing the front and back of the robot to operate independently. The bladders are confined within a fabric sleeve.  This sleeve not only limits radial expansion of the bladders but also acts as soft armor, protecting the robot from environmental hazards that could cut or puncture the bladders.  The sleeve is secured to the bladders using tie wraps on either end.

Side view of inflated robot.

Top and front view of uninflated FRCMR.

The robot has feet located at either end of the sleeve. Each foot is a fabric patch that features multiple rows of bristles. Each bristle has bare fabric on the forward facing side and a rubber coating on the rear facing side. Using directional friction, the robot exhibits forward locomotion through linear expansion and contraction. The foot patches can be removed or replaced as required.

Through different inflation patterns, several modes of locomotion can be achieved.

Crawling Motion

Crawling Motion

The simplest mode of locomotion occurs when all three bladders extend and retract simultaneously. This produces a highly stable flat earthworm like crawling motion.

Inchworm Motion

Inchworm Motion

While performing the inchworm motion the robot bends in an upside down U and then straightens.   This is more air-efficient because the bladders only partially deflate. It is also the least stable mode of transport.   

Turning Motion

Turning Motion

The turning motion occurs when the bladders on one side of the sleeve inflate, allowing the robot to pitch in the opposite direction.

Climbing Motion

Climbing Motion

During the climbing motion, the front of the robot bends upwards, extends and then bends back down on top of the obstacle .  

Controls

The robot is controlled via a fluidic control board detailed on the Soft Robotics Toolkit using a closed feedback loop that reads the pressure of each bladder.

 

Fabric-Reinforced Actuator

Initially, the body of the robot was made of an extending fiber-reinforced actuator. The robot was able to achieve reliable forward motion. However, when the robot extended, bubbles of rubber appeared in the gaps between wrappings. The example pictured was due to inconsistent winding; however, it highlights a weakness in the design of extending fiber-reinforced actuators. These actuators have a finite number of wrappings; as the actuator extends, the gap between wrappings increases. Eventually the internal air pressure forces sections of the bladder between the gaps in the wrappings. This problem can be mitigated by wrapping the bladder evenly and having a uniform wall thickness. However, the tolerances required are too small to be kept reliably. 

The first prototype to address this problem surrounded the bladder with fabric before wrapping it. The intent was that bubbles that normally would form between wrappings would be suppressed by the fabric. This was moderately successful. The bubbles were partially suppressed by the fabric. However, the actuator was extremely difficult to make because the sleeve would unravel as the actuator was being wrapped.

In the next iteration, the fabric was stapled together so that it formed a sleeve. As this new actuator was being wrapped, it was realized that the sleeve made the wrapping obsolete and the wrapping was removed. The sleeve worked well, the bladder inflated evenly with no bubbles or obvious weak points.

The next iteration used the same fabric sleeve; however, it was sewn together rather than stapled. This made the sleeve stronger and reduced the risk of puncture. This actuator defied all expectations. Since the sleeve reinforced the bladder continuously, no bubbles formed at all.  It was decided to call the new actuator a fabric-reinforced actuator.

Fabric-reinforced actuators have many advantages over fiber-reinforced actuators.

Stronger: By reinforcing the bladder with a sleeve, the robot becomes significantly more durable. With no wrappings, the bladder walls can be thinner. Due to the physical constraints, the bladder is not capable of bubbling through the sleeve. 

During prototyping, a 5“ long actuator with 1/16 inch thick walls was able to extend to 13 inches. At this length the actuator could still bend 180 degrees and suffer multiple impacts with the head of a hammer without taking any observable damage. A traditional fiber-reinforced actuator produced multiple bubbles and popped in similar circumstances.

Protection:

 

Soft robots by nature are resistant to blunt trauma. However, they are extremely vulnerable to cuts and punctures, especially when inflated. The fabric sleeve can act as a soft suit of armor that protects the robot from environmental hazards. The sleeve pictured is a made from Kevlar, and is used for arm protection. When used for an extending actuator, the sleeve was able to protect the bladder from repeated cuts with a meat knife. Even fabrics like cotton effectively protect the robot from cuts.

Variability:

Both sleeves are inflated by the same bladder. (right) Unmodified sleeve. (left) Sleeve with an inextensible layer of hot glue.

Many soft robots undergo some form of mechanical programming. Examples include the addition of an inextensible layer or varying of the wrapping angle. One drawback of these methods is that they are internal. Once done, it is nearly impossible to undo. For example, if an actuator is internally programmed to curl via the addition of an inextensible layer, it cannot be reprogrammed to extend. By externally modifying the sleeve, the fabric sleeve can be used to program actuators externally. A robot designed to curl can be reprogrammed to extend by simply changing sleeves. Most sleeves can be programmed in a fraction of the time it would take to make a new actuator. Sleeves are also reusable and can be kept on hand for prototyping or new designs. This has the potential to make prototyping new robots significantly faster, easier and more economical.  Additionally, fabric-reinforced actuators are not limited to single worm forms. Future robots can take any shape.

Feet

Robot Evolution

The feet enable the robot to turn reciprocating motion into forward motion.  

Successful feet have two qualities:

  • They must allow the robot to move forward.
  • They cannot inhibit the soft nature of the robot.

Foot Prototypes

All of the foot prototypes were designed to work the same way. They have two contact surfaces, a high friction rubber coated surface and a low friction rubberless surface. The feet achieve directional friction by pivoting between these two surfaces. When the foot moves forward, the low friction surface slides across the ground uninhibited. As the robot moves the foot backward, the high friction surface makes contact with the ground and inhibits movement, the rest of the robot is then moved forward.

(left) Cardboard sled leaning on low friction side. (right) Upside down image of cardboard sled showcasing rubber pads.

The proof of concept prototype was a cardboard sled. In this prototype, each end of the robot was outfitted with a cardboard sled. The high friction surface was the rubber pads  (seen in the image on the right) and the low friction surface was the sloped cardboard.  When the sled was dragged forward, the entire sled pivoted forward so that only the bare cardboard touched the ground.  When pushed back the sled pivoted and the rubber pads made contact with the floor. This prototype allowed the robot to move forward reliably and confirmed viability of this concept. However, the sleds were bulky and too cumbersome to be used.

The second foot prototype was designed to be a smaller less obstructive version of the cardboard sled. Like the cardboard sled, it was designed to switch between a rubber coated high friction surface and a low friction surface. These sleds were 3D printed and designed to go underneath the robot. The sleds were connected to the body of the robot with a piece of fabric that was glued to the body of the worm. This prototype failed because the feet did not pivot reliably. Some feet remained bent in the low friction position despite not being in contact with the ground (see 2nd foot from the left). It was also noticed that the rigid feet could not conform to the body of the inflated robot. Furthermore, it was thought that the stiff extrusions on the bottom of the robot would inhibit motion if the robot was to crawl over debris.

The next foot prototype was built out of the same type of fabric used to make the sleeve. It was a skirt that had a coating of rubber on the underside. Bristles were cut into the skirt to allow individual sections to pivot. The foot was ineffective because the rubber bled through the fabric and resulted in little differentiation between the high and low friction surfaces. However, this design confirmed that it was possible to create a foot that conformed to the shape of the robot.

The next iteration was made with two pieces of fabric glued together using hemming tape. The tape formed a barrier that the rubber could not permeate.  While this kept the frictional differentials intact, it also made the fabric too stiff to pivot between the high and low friction surfaces. To solve this issue, the fabric was bent away from the robot and the bristles were shortened to allow for easier pivoting. This decreased the angular displacement needed to pivot between states and lifted the robot up so that all of the weight was on the feet. The robot moved forward reliably and the foot pads were soft enough to conform to the shape of the inflated robot. This prototype was successful.

Foot Placement

The feet were initially evenly spaced along the robot. The intention was that a larger number of feet would equate to more traction. It was thought that all of the feet would contribute to the forward movement. However, during testing, all of the movement was focused around two individual feet, one in the front and one in the back.    

The next iteration used less feet and placed them on the front and back thirds of the robot. Again, the resulting movement was centered around two feet.

To save time, a sled was developed to test a new foot placement on the existing sleeve. The feet were sewn to a piece of fabric that was then wrapped around a block of rubber and butterfly clipped to the ends of the sleeve. This prototype was successful, a large amount of expansion was converted to forward motion. Unfortunately, the sled was too heavy to be viable.

Next, the same foot placement was used; however, the feet were sewn directly onto the sleeve. This was very successful.

Throughout testing, it appeared that one primary foot on each side of the robot did most of the work. It was decided that 2-3 rows of bristles per foot allowed for reliable forward movement.

Up until this point the feet were sewn directly onto the sleeve. However, this meant that the feet had to be sewn onto the fabric before the sleeve was made. To simplify the manufacturing process going forward, the feet were pre-assembled on fabric patches that could be attached via hemming tape or hot glue after the sleeve was completed.

The ability to attach patches post production increases the potential applications for fabric-reinforced actuators exponentially. In this project the patches were only used to attach the feet; however, in future development, patches can be used to connect small instruments such as sensors, thermometers, and cameras to the exterior of the robot. Therefore, on-site modifications can be made once needs are identified, allowing for individually tailored solutions to unique problems.

The design is not perfect, dirt and dust stick to the rubber side and decrease its surface friction. Dirty feet were still able to move the robot forward; however, the movement was noticeablely less efficient compared to a clean foot. Additionally the feet were designed to move over smooth ground. This currently limits the environments that the robot would be effective in.  Building versatile feet was reserved for a future enhancement. 

Sleeve

The sleeve gives the robot structure and is created by sewing a tube out of an inextensible cotton fabric. It is connected to the front and back ends of the bladders with zip ties. When the bladders are not inflated, the sleeve is compressed longitudinally (similar to pushing up the sleeve of a long sleeve shirt). As the bladder inflates, it expands, filling the sleeve. In this project all of the sleeves are designed for extension. 

In most soft robots, a single puncture in one of the bladders indicates the entire robot would need to be replaced. However, in the fabric-reinforced actuator, a punctured bladder can be easily removed and replaced in a few minutes with no special equipment required.  This makes the robot not only completely soft but also repairable in the field.

(left) Single cavity sleeve and bladder. (right) Extended single cavity sleeve.

The first sleeve prototype was a single cavity sleeve. It was made from a single piece of fabric that was curled into a tube. This meant that the sleeve needed to be sewn by hand because a sewing machine would have sewn through both sides of the tube. Functionally, this sleeve had a tendency to roll sideways when inflated and had no turning capabilities. However, it did work well enough to test the fabric-reinforced actuator concept. Once the concept was proven viable, it was used to test the various foot prototypes.

(top left) Sleeve pattern. (top right) One side inflated. (below) Sleeve with both bladders inflated.

The second sleeve prototype was designed with two side by side bladder compartments separated by one half inch of fabric. It was made from two pieces of fabric that were sewn together using a sewing machine. This method made it easier to produce multi bladder sleeves. Additional compartments could be made by sewing more lines into the fabric.  Functionally, the wider sleeve was more stable than the first prototype. Additionally, the arrangement of the two bladders made it possible to turn left or right by inflating one side of the sleeve at a time. However, it was unable to bend up (a movement necessary for climbing). This sleeve was used to test and determine foot placement.

Side view of inflated robot.

The final sleeve prototype added a third bladder compartment sewn on to the top of the previous prototype design.  This gave the robot the ability to bend up and down in addition to left and right.  The top compartment is only attached to the rest of the robot at both ends and roughly an inch in the middle of the sleeve. This allowed the robot to bend upwards at a corner rather than a curve and made it easier to climb over objects.  This sleeve was used in the final design.

The folds in the compressed sleeve highlight a potential design weakness. These ripples are more likely to get caught on objects than the smooth rubber wall. Different sleeve materials can be chosen to minimize snagging.

Bladders

The robot is actuated by inflating silicone bladders.  All of the bladders are hollow rectangles made of Ecoflex 00-30. Rectangular bladders were chosen because they were the simplest to produce. Early bladders were made using the cardboard mold process from the Soft Robotics Toolkit. Later bladder molds were fabricated on a 3D printer. After the hollow cavity cured, the mold was placed in a shallow pool of rubber to complete the bladder.
Cardboard Mold
First bladder prototype mold.

The first bladder prototype was a simple hollow rectangle made using a cardboard mold. The walls of the bladders were about ¼ of an inch thick.  These were used in the early development of the fabric-reinforced actuator.

The eventual goal is for this type of robot is to be untethered. This required that both the compressor and the power source would need to fit on the robot. Bladders with thin walls require less air pressure to fully inflate. Using thin walls would mean that smaller, less powerful compressors and battery sources would be needed to run the robot.  

Mold for second bladder prototype. 3/32 inch wall thickness.

The remaining bladder mold prototypes were fabricated on a 3D printer. Three wall thickness were tested: ⅛ inch, 3/32 inch, and 1/16 inch. All three thickness were able to inflate the sleeve without sustaining damage. However, since this prototype was tethered, they all exhausted at different rates. The 3/32 inch wall was chosen because it was a balance between thin walls and a workable exhaust rate.

The bladders are tied to the sleeve with a zip tie. However, this turned out to be a design flaw. When the bladder inflated, the edges of the zip tie would cut into the bladder and eventually pop it. The solution was to increase the wall thickness on the each end of the bladder so that the zip tie was farther away from the inflated section of the bladder. Notches were added to the design to for zip tie placement.

Final Bladder Mold
Final Bladder Mold

When attempting to get the robot to climb, it became apparent that it would be beneficial to have the front and back of the robot extend independently. To do this, the bladder was divided in two by a ¼ inch wall of rubber. This effectively created 2 separate bladders and allowed the front and back halves of the robot to be operated independently.   

The bladders prototyped are not designed specifically to expand linearly. They expand in all directions and the sleeve limits the radial expansion.  While this is effective and acceptable, it is not efficient. If the bladder walls folded like an accordion, the bladder would already have a bias towards linear expansion and less energy would be wasted expanding radially. This is a consideration for the future versions.

Motion

The Fabric-Reinforced Controlled Motion Robot (FRCMR) moves by inflating sections of its body. The diagram shows the inflation patterns used to achieve various modes of locomotion. Each section of the diagram represents a bladder on the robot.  Bladders are inflated to 4 PSI.

 

Crawling Motion

Inchworm Motion

Turning Right

Turning Left

 

Climbing Motion

 

Control

Complete control board setup.
The Fabric-Reinforced Controlled Motion Robot (FRCMR) is controlled using the fluidic control board found on the Soft Robotics Toolkit. Functionally, the board performs the same tasks; however, some parts were replaced in order to maintain affordability. The pump was replaced with a Good Year 120 Volt Multi-Purpose Inflator due to the significantly higher flow rate.

Robot Configuration

The final prototype of the robot used 6 bladders. This was a problem because the control board was designed with 4 outputs. Modifying the control board to have more outputs was not economical. To work around this, two outputs were shared between 4 bladders. However, this meant that the robot could not be operated to its fullest capabilities without changing the tube configurations. All of the functions of the robot are split between two configurations.  

Configuration 1 connects the two halves of each bottom bladder. This can be done by either connecting the front and back of the robot with a Y fitting or by using the second prototype bladders.  As a result, in this configuration the front and back of those bladders cannot be operated independently.  

Configuration 2 connects the two front and two back halves of each bladder. This is best done with Y fittings.


Each color is connected to a specific controller output.

A potential upgrade to the control panel is to add outputs to the board. If the board had 6 total outputs, the different configurations would be obsolete, and FRCMR would be able to go from turning to climbing through simple bladder inflation changes.

Robot Controls

Pot 1Pot 2 Pot 3 Pot 4
Switch 1Switch 2Switch 3Switch 4
 

Control setup 1: Pots 3 and 4 turned all the way to the left. Robot must be set in configuration 1.

Pressure ControlUnusedMinMin
Crawling MotionRight TurnLeft TurnUnused
 

Control setup 2: Pot 3 turned all the way to the right.Robot must be set in configuration 1.

Pressure ControlUnusedMaxMin
Walking MotionUnusedUnusedUnused
 

Control setup 3: Pot 4 turned all the way to the right. Robot must be set in configuration 2. Used for manual control.

Pressure ControlUnusedMinMax
Red BladdersOrange BladdersBlue BladderGreen Bladder
 

Pressure sensors: The programming for the robot is very simple. The robot uses a bang bang control to regulate pressure in the actuators. If the pressure in the bladder is lower than the desired pressure, the valve is opened. If the pressure in the bladder is larger than the desired pressure, the valve is closed.

The pressure sensor is located next to the manifold. When the valve opens, a wave of high pressure air travels through the tube. Due to the sensors' location, the high pressure wave was detected before it reached the bladder. This resulted in high pressure readings that closed the valve prematurely. To mitigate this, the tee fitting was relocated as close to the robot as possible. This resulted in much more accurate pressure readings.

Control Board Accessories

The fluidic control board has a few enhancements that made it easier to fabricate and operate the Fabric-Reinforced Controlled Motion Robot (FRCMR).  

Degassing chamber

An affordable degassing chamber was made out of a used glass jar with lid. The pump used in the fluidic control board is also a vacuum pump. 

Air Tanks

FRCMR requires a lot of air. For smooth, continuous movement it was best to store air in external air tanks. Air tanks were made from 2 liter soda bottles. Tubing was inserted into the cap and connected to the pump and manifold by a Y connector.
In addition to passive air tanks, the bottles can be connected directly to the robot acting as two liter squeeze bulbs. Most of the prototypes were tested using this method of inflation.