Article FludoJelly: Experimental Study on Jellyfish-Like Soft Enabled by Soft Pneumatic Composite (SPC)

Aniket Joshi, Adwait Kulkarni and Yonas Tadesse *

Humanoid, and Smart systems (HBS) Laboratory, Mechanical Engineering Department, The University of Texas at Dallas, Richardson, TX 75080-3021, USA * Correspondence: [email protected]; Tel.: +972-883-4556

 Received: 1 June 2019; Accepted: 11 July 2019; Published: 15 July 2019 

Abstract: Several bio-inspired underwater have been demonstrated in the last few years that can horizontally swim using different smart actuators. However, very few works have been presented on robots which can swim vertically, have a payload and resemble a jellyfish-like creature. In this work, we present the design, fabrication, and performance characterization of a new tethered robotic jellyfish, which is based on inflatable soft pneumatic composite (SPC) actuators. These soft actuators use compressed air to expand and contract, which help the robot to swim vertically in water. The soft actuators consist of elastomeric air chambers and very thin steel springs, which contribute to gaining faster motion of the biomimetic robot. A prototype of 220 mm in diameter and consisting of eight actuating units was fabricated and tested underwater in a fish tank. It reached a height of 400 mm within 2.5 s while carrying a dead weight of 100 g when tested at 70 psi (483 kPa) pressure. This high performance (160 mm/s on average speed) suggests that faster motion with a payload can be achieved by using SPC actuators. The inflatable structures help to flap the bell segments as well as in buoyancy effect for rapid vertical motion. The major achievement of this work is the ability to demonstrate a novel use of inflatable structures and biomimetic flapping wings for fast motion in water. The experimental and deduced data from this work can be used for the design of future small unmanned underwater vehicles (UUVs). This work adds a new robot to the design space of biomimetic jellyfish-like soft robots. Such kind of vehicle design might also be useful for transporting objects underwater effectively.

Keywords: biomimetic robots; jellyfish; artificial muscles; artificial life; swimming robot; soft robot

1. Introduction Biomimetics implies the study of anatomy and mechanisms present in nature and creating a model or system using synthetic materials for solving complex engineering problems. This approach is crucial for advancements in science and technology [1]. is one such field, which focuses on replicating the movement of biological creatures using soft synthetic materials. Compared to soft robots, conventional robots have rigid structures that limit their motions. Soft robotics brings in a new dimension of motion due to soft robots’ fluidic movement and many degrees of freedom [2–4]. Pneumatic networks (Pneu-Nets) are fluidic actuators, which bend with the help of compressed air [5–7], their motion direction can be controlled by changing the design and placement of the inextensible layer. The wall thickness and the size of the chamber affect the motion and the time required to inflate it completely [8,9]. Based on this, many biomimicry related studies and inventions have been made, such as a soft robotic tongue [10], soft robotic manipulators [11], a quadruped [12], an octobot (a soft robot with microfluidic controller) [13], a snake-like robot and a self-contained robotic fish [14]. Soft robotics is not only limited to biomimicry, but also has been extended to wearable devices to help the

Robotics 2019, 8, 56; doi:10.3390/robotics8030056 www.mdpi.com/journal/robotics Robotics 2019, 8, 56 2 of 20 rehabilitation of physically impaired individuals. Thumb assistance and ankle foot rehabilitation are a few examples of soft wearable devices [15]. Most of the biomimetic underwater robots presented in the literature are based on the common fishes (Tuna, carp, and catfish). However, jellyfish species have been considered by researchers in recent years. Based on the swimming mechanism, jellyfish can be categorized into two types, called Prolate and Oblate. Prolate species use jetting type and Oblate species use rowing type of propulsion [16]. Jellyfish of different species have been chosen for biomimicry due to their energy efficiency during swimming, the ability to incorporate payload and the ability to select various sizes for a model vehicle since there are species that range from millimeter to meter scale. Biologist explained that the motion of a jellyfish is due to the contraction of the bell and expulsion of fluid through the bell to generate a thrust [16]. The movement of the bell and the flow pattern produces unsteady vortices, which are also highly responsible for its motion. Previously developed jellyfish-like robots have used shape memory alloy (SMA), ionic polymer metal composite (IPMC) and electrical motors to mimic the bending motion. An example of jellyfish developed using electric motor and linkage mechanisms is AquaJellies 2.0 developed by Festo Corp., which has elegant gliding motion through water [17]. Xiao et al. showed the implementation of a jellyfish-like motion using servos and four six-bar linkage mechanism [18]. Villanueva et al. in 2009, presented a jellyfish named JetSum made of thin aluminum sheets actuated using biometal fiber (a type of SMA). The jellyfish was able to mimic the motion of a natural jellyfish with jetting type motion, the main challenge for that work was the high-power consumption for the multiple SMAs [19]. A more biomimetic version, named Robojelly was introduced by Villanueva et al. in 2011 [20]. Powering aspect has been investigated by Tadesse et al., in 2012 using composite actuators consisting of multiwall carbon nanotubes, shape memory alloy, and platinum particles that enable the utilization of fuel power such as hydrogen and oxygen gas [21]. Another large-size jellyfish inspired by cyanea capillata species was developed using linear actuators in 2013 by Villanueva et al. [22]. In the extremely small size, Guo et al. used IPMCs to develop a microrobot jellyfish. The system was efficient in creating thrust but did not have the ability to carry payload [23]. Naworth et al. showed a novel jellyfish created using chemically disassociated rat tissue and silicone polymer, which responded to electrical field stimulation. They discussed the use of natural material and tissue to create the jellyfish that was never done before [24]. However, the approach uses several chemicals and may be challenging to implement in macroscale robotic systems. Recently, Frame et al., in 2018 showed a soft jellyfish robot that uses hydraulic systems (water based) for actuation using DC motor pumps. It was demonstrated the robot performed successful field test in the ocean [25]. In our robot, we used a pneumatic system using compressed air and the geometry of the tentacles are also different, they used oblong shape to compensate unequal flow of water in their tentacle design. However, we used a single compartment design for our prototype presented in this study. The principal movement mechanism in jellyfish-like robots is thrust generation by jetting motion of the fluid beneath the robot. The objective of this paper is to investigate the performance of a jellyfish-like robot such as vertical movement in a fish tank, cyclic motion and deformation with and without payload at different heights in a water column. Payload and swimming ability are important characteristics of such robots when it comes to moving an object from point A to B in water. In this work, we used soft pneumatic composite (SPC) actuators to produce the contraction and expansion of the bell. These are completely different from the SMAs [19], SMA composite (BISMAC) [26], servos [22] and conducting polymers [27] which our group and collaborators used before. We called the prototype FludoJelly and it is powered by an air compressor, which inflates the actuators and bends them to create a smooth motion. The actuation required a quick pulse input followed by recovery phase. The thrust was generated in the quick actuation phase followed by a gradual relaxation phase. Position and velocity were measured under different input conditions. Sung-Weon et al. discussed a similar bio-inspired pulse for the actuation of their IPMC actuated jellyfish [28]. The other interesting aspect of this FludoJelly is the “inflatability” of the structure. Inflatable structures are found in various creatures for various purposes. Figure1A–C shows various jellyfish species, moon jellyfish ( Aurelia aurita) that Robotics 2019, 8, x FOR PEER REVIEW 3 of 20

the sizes range from 100 to 150 mm. This animal resembles a jellyfish, but it is not a single multicellular organism (like the common jellyfish). It is a colony of small specialized animals. It has an inflated structure at the top that enables it to float keeping the rest of the body to be under the water. The species shown in Figure 1B and 1C do not have inflatable structures, rather they use an undulatory motion of their bell for movement. Figure 1D–1F show other animals with inflatable structures in their body. A frog is a typical example that inflates its vocal sac to attract mates. Another example is a frigate bird which has a wingspan of 2.3 m, and also other marine mammals such as a Roboticswalrus2019, 8 ,which 56 uses inflation for buoyancy and to sleep better [29]. 3 of 20 Considering the benefits of using soft composite and inflatable structures, we have designed and developed the novel fluidic actuated robot, FludoJelly. This robot has a faster speed than the existing rangespr fromototypes 100 while to 400 carrying mm, and a payload, true jellyfish, which has Mastigiidae not been demonstrated (Phyllorhiza in) that any rangesliterature from so far. 300 The to core 600 mm. Figurecontribution1A is a colony of this of work jellyfish-like is the data creatures obtained from also knownthe prototype as Portuguese using the experimental man-of-war method, ( Physalia the) and the sizesfabrication range from techniques, 100 to and150 mm. the design. This animal This data resembles helps in a creating jellyfish, such but a it fast is not-moving a single jellyfish multicellular-like robot. The displacement of the robot while carrying a payload, the influence of pressure on the organism (like the common jellyfish). It is a colony of small specialized animals. It has an inflated deformation of the bell segment at different heights in a water column, and cyclic motion of the robot structure at the top that enables it to float keeping the rest of the body to be under the water. The while tethered are some of the key contributions. species shownThe paper in Figure is organized1B,C do as not follows. have First, inflatable we will structures, describe the rather design they of a use single an undulatoryactuator and motionthe of theirfabrication bell for movement.of the jellyfish Figure prototype;1D–F showwe then other describe animals the withexperimental inflatable setup structures and data in collection. their body. A frog isResults a typical and example discussion that will inflates be the its next, vocal followed sac to attractby the swimmingmates. Another test and example comparison is a frigate of the bird whichswimming has a wingspan performance of 2.3 ofm, our and robot also other with existing marine prototypes.mammals such Conclusion as a walrus and future which works uses inflation are for buoyancypresented and at the to end. sleep better [29].

FigureFigure 1. Jellyfish 1. Jellyfish species species and and animals animals that that have have inflatable inflatable structures structures.. (A ()A Portuguese) Portuguese man man-of-war,-of-war, physalia,physalia,courtesy courtesy of Pixabay of Pixabay GmbH GmbH/Alicia/Alicia Campbell; Campbell; (B)) Moon Moon Jellyfish, Jellyfish, AureliaAurelia aurita aurita, photo, photo courtesy courtesy of Armitaof Armita Hamidi; Hamidi (C); ( TrueC) True Jellyfish, Jellyfish, Mastigia, Mastigia, photo photo courtesy of of Quin Quin Marshall Marshall [30 []30; (D];) (Frog,D) Frog, LitoriaLitoria chloris [31]; (E) Frigatebirds, courtesy of Galapagos Conservation Trust, photographer Martin Partridge; chloris [31]; (E) Frigatebirds, courtesy of Galapagos Conservation Trust, photographer Martin Partridge; (F) Walrus, Odobenus rosmarus, photo courtesy of Pixabay GmbH/Sven Schlieter. (F) Walrus, Odobenus rosmarus, photo courtesy of Pixabay GmbH/Sven Schlieter. 2. Materials and Method Considering the benefits of using soft composite and inflatable structures, we have designed and developed2.1. Soft Pneumatic the novel Composite fluidic (SPC) actuated Actuator robot, FludoJelly. This robot has a faster speed than the existing prototypes while carrying a payload, which has not been demonstrated in any literature so The pneumatic actuator is based on the actuator design originally developed by Harvard far. The core contribution of this work is the data obtained from the prototype using the experimental University [32]. However, our actuator consists of one single chamber and thin spring steel in the outer method,layer, the which fabrication inflates to techniques, create a bending and motion. the design. To create This the data bending helps motion in creating an inextensible such a fast-movingmaterial jellyfish-likeis embedded robot. into The the displacement soft material. Paper of the and robot PDMS while (polydimethylsiloxane) carrying a payload, are thecommonly influence used of as pressure an on theinextensible deformation layer of in the soft bell robots. segment The main at di ideafferent behind heights designing in a water the actuator column, was and to have cyclic a maximum motion of the robotthicknes while tethereds on the sideare someand minimum of the key thickness contributions. on the top to obtain the maximum utilization of the The paper is organized as follows. First, we will describe the design of a single actuator and the fabrication of the jellyfish prototype; we then describe the experimental setup and data collection. Results and discussion will be the next, followed by the swimming test and comparison of the swimming performance of our robot with existing prototypes. Conclusion and future works are presented at the end.

2. Materials and Method

2.1. Soft Pneumatic Composite (SPC) Actuator The pneumatic actuator is based on the actuator design originally developed by Harvard University [32]. However, our actuator consists of one single chamber and thin spring steel in the outer layer, which inflates to create a bending motion. To create the bending motion an inextensible material is embedded into the soft material. Paper and PDMS (polydimethylsiloxane) are commonly used as an Robotics 2019, 8, 56 4 of 20 inextensible layer in soft robots. The main idea behind designing the actuator was to have a maximum thickness on the side and minimum thickness on the top to obtain the maximum utilization of the amount of air entering the chamber. The pressure inside the chamber will follow the path of least resistance, thereby inflating more from the top layer of the actuator compared to the sides. Figure2A showsRobotics the 2019 cross, 8, x FOR section PEER andREVIEW the side view of the actuator, which was made of two-piece mold5 of 20 and casting silicone.

Figure 2. Single soft pneumatic composite (SPC) actuator. (A) Schematic diagram of the actuator fromFigure two 2. views; Single (softB) Actuation pneumatic at composite different ( pressureSPC) actuator. (i) 0 psi(A) beforeSchematic actuation, diagram (ii) of atthe 2 psi,actuator (iii) atfrom 7 psi; two views; (B) Actuation at different pressure (i) 0 psi before actuation, (ii) at 2 psi, (iii) at 7 psi; (C) Pulsed actuation (i) at 0 psi 0 ms pulse duration, (ii) at 10 psi 100 ms pulse duration and (iii) at 10 psi (C) Pulsed actuation (i) at 0 psi 0 ms pulse duration, (ii) at 10 psi 100 ms pulse duration and (iii) at at 250 ms. The actuator geometry is 76 mm 19 mm 15 mm in length (L1), width (w1) thickness (h). 10 psi at 250 ms. The actuator geometry× is 76× mm × 19 mm × 15 mm in × length (L1), The other dimensions are t1 = 2 mm, t2 = 3 mm, t4 = 5mm, t5 = 127 µm, w2 = 12.7 mm, and w3 = 21 mm; width (w1) × thickness (h). The other dimensions are t1 = 2 mm, t2 = 3 mm, t4 = 5mm, t5 = 127 µm, (D) Vertical movement of SPC actuator vs time. w2 = 12.7 mm, and w3 = 21 mm; (D) Vertical movement of SPC actuator vs time.

2.2. Biomimetic Jellyfish Design The design of the robotic structure is inspired from movement of the natural jellyfish similar in overall structure as our previous prototypes. It has eight SPC actuators arranged radially like a disc. The circular array of actuators is cast from a two-piece mold as shown in Figure 3A,B.

Robotics 2019, 8, 56 5 of 20

The silicone used for our actuator is Ecoflex-30, which has a shore hardness of 00–30, up to 900% elongation at break, and a tensile strength of 200 psi (1.4 MPa) [33]. The new structure is the use of spring steel (blue rectangle in Figure2A) as the inextensible layer at the bottom. The advantage of using spring steel is its spring-back effect. When a force is applied to a spring steel on one end and released suddenly, it retracts to its original position quickly. This structure was effective for SMA based systems as well [20]. In our case, the force is applied when compressed air is filled in the chambers. The material of the spring is 1095 blue tempered steel with a thickness of 0.005” (t5 = 127 µm) and a width of 0.5’ (w2 = 12.7 mm). The leaf spring is embedded in a skin silicone. It takes 3–4 hours for the silicone to fully cure under room temperature. A tube for inflation is inserted into the actuator through a nipple extrusion and zip tie is wrapped around the nipple to hold the inserted tube in place. Figure2B shows the undeformed SPC actuator at the first column followed by the deformed shapes of the actuator at pressure 2 and 7 psi (14 and 48 kPa). Figure2C shows the inflated structure at different actuation times 0, 0.1 and 0.25 s (video S1, supplementary file, Video S1 FludoJelly1, SPC actuators.mp4). The vertical displacement is 22 mm as shown in the time domain plot in Figure2D, the left axis. When normalized with the length, the maximum vertical displacement is 28% as shown in the right axis. The negative indicates the direction of bending. The bending angle (θ) of the actuator is a function of three parameters: geometry, input pressure and material constants such as elasticity. This relation can be described by the equation obtained from Onal et al. [34]. ! 1 L ε(σ) θ = 2n tan− (1) 2h L represents the length of the channel of the actuator and h denotes the height of the actuator channel, ε(σ) is the strain of the actuator which is a nonlinear function of stress σ, n is the number of channels for multi-chambered actuator, in our case n = 1. Additionally,

h σ = P (2) ∆h where σ is the stress developed due to internal pressure P supplied to the actuator and ∆h is the change in height due to pressure of the actuator. From the above two equations (Equations (1) and (2)), the bending angle of such actuator is dependent on pressure, geometry of the actuator, elastic constants in a nonlinear manner [35,36]. The exact equation that predict the deformation is more complicated as the actuator consists of multiple materials. However, the simplified equations will provide an insight on the behavior of the actuator and guide the research. Our focus in this paper is mainly on experimental investigation of the jellyfish-like robot. We do not provide rigorous equations and validation with experimental data due to the complex nature of the problem.

2.2. Biomimetic Jellyfish Design The design of the robotic structure is inspired from movement of the natural jellyfish similar in overall structure as our previous prototypes. It has eight SPC actuators arranged radially like a disc. The circular array of actuators is cast from a two-piece mold as shown in Figure3A,B. The leaf-springs are arranged in the same orientation as the actuators (Figure3C). This approach improves the swimming speed. The cast created earlier is then bonded with the silicone skin embedded with spring steel seen in Figure3D. The thickness of the skin is 3 mm. The main aim of the prototype is to mimic the bell contraction and expansion movement to gain lift. Air is supplied through a single tube through the center portion from an air compressor. The actuators bend downwards creating a contracted bell like structure. Figure3E,F shows the actuation before and after respectively. One of the challenges faced in fabricating the structure was bonding the layers and sealing it. Depending on the thickness of the air chamber, the trapped air during casting can give rise to small air bubbles in the structure. These air bubbles can create pores causing leaks inside the air chamber. Such problems can be avoided by casting silicone inside a vacuum chamber. Robotics 2019, 8, x FOR PEER REVIEW 6 of 20

The leaf-springs are arranged in the same orientation as the actuators (Figure 3C). This approach improves the swimming speed. The cast created earlier is then bonded with the silicone skin embedded with spring steel seen in Figure 3D. The thickness of the skin is 3 mm. The main aim of the prototype is to mimic the bell contraction and expansion movement to gain lift. Air is supplied through a single tube through the center portion from an air compressor. The actuators bend downwards creating a contracted bell like structure. Figure 3E,F shows the actuation before and after respectively. One of the challenges faced in fabricating the structure was bonding the layers and sealing it. Depending on the thickness of the air chamber, the trapped air during casting can give rise to small air bubbles in the Roboticsstructure.2019 ,These8, 56 air bubbles can create pores causing leaks inside the air chamber. Such problems6 ofcan 20 be avoided by casting silicone inside a vacuum chamber.

FigureFigure 3. 3. JellyfishJellyfish inspired inspired robot robot based based on on fluidic fluidic actuation actuation and and spring steels—FludoJelly.steels—FludoJelly. ((AA)) Computer-aidedComputer-aided designdesign (CAD)(CAD) modelmodel ofof thethe 2-piece2-piece moldmold forfor castingcasting thethe softsoft roboticrobotic jellyfishjellyfish structure;structure; ((BB)) elastomericelastomeric chamberschambers created byby thethe mold;mold; ((CC)) 10951095 leafleaf springspring embeddedembedded in silicone;silicone; ((DD)) ElastomericElastomeric chamber chamber and and leaf leaf spring spring embedded embedded together together to form to the form bell; the (E bell) Jellyfish; (E) Jellyfish in its deflated in its phasedeflated or relaxationphase or relaxation phase; (F )phase its inflated; (F) its stage inflated or contraction stage or contraction phase. phase.

2.3.2.3. Experimental Setup for Testing SwimmingSwimming TheThe swimmingswimming motionmotion ofof thethe robotic jellyfishjellyfish waswas tested using didifferentfferent camerascameras whilewhile thethe robotrobot waswas tetheredtethered underwaterunderwater asas shownshown inin thethe schematicschematic diagramdiagram inin FigureFigure4 4AA.. These These cameras cameras were were used used forfor characterizingcharacterizing andand verifyingverifying thethe motionsmotions with didifferentfferent measurementmeasurement conditionsconditions andand atat didifferentfferent frameframe rates.rates. A pressurepressure controlcontrol circuitcircuit waswas mademade toto controlcontrol thethe actuatorsactuators asas shownshown inin thethe schematicschematic diagramdiagram inin FigureFigure4B 4B and and the the photograph photograph of theof the setup setup is shown is shown in Figure in Fig4D.ure The 4D air. The compressor air compressor used here is single stage air compressor (California Air tools Model no. 202977393) which can generate a maximum pressure of 120 psi. Solenoid valves (STC 2P025) were used to direct the air flow in the pneumatic circuit, the valves were 2/2 valves and required a 12 V power supply. An external DC power source was used to provide the required voltage. A MOSFET was used to act as a switch and energizing the solenoid. A pressure sensor (Honeywell-ASDX AVX030PG2A5) was also installed in the circuit for pressure readings. The data from the pressure sensor was collected from a National Instruments data acquisition unit NI USB 8452 through LabVIEW program. The pressure sensor used here is an I2C network interface digital output sensor. Robotics 2019, 8, x FOR PEER REVIEW 7 of 20 used here is single stage air compressor (California Air tools Model no. 202977393) which can generate a maximum pressure of 120 psi. Solenoid valves (STC 2P025) were used to direct the air flow in the pneumatic circuit, the valves were 2/2 valves and required a 12 V power supply. An external DC power source was used to provide the required voltage. A MOSFET was used to act as a switch and energizing the solenoid. A pressure sensor (Honeywell-ASDX AVX030PG2A5) was also installed in the circuit for pressure readings. The data from the pressure sensor was collected from a National

InstrumentsRobotics 2019, 8 ,data 56 acquisition unit NI USB 8452 through LabVIEW program. The pressure sensor used7 of 20 here is an I2C network interface digital output sensor.

(A) (B)

(C) (E)

(D)

FigureFigure 4. 4. ExperimentalExperimental setup setup for for testing testing (A) Position (A) Position and velocity and velocity tracking tracking of the FludoJelly; of the FludoJelly (B) Circuit; (forB) Circuit actuation for actuation test; (C) Blocking test; (C) Blocking force determination force determination of the SPC of the actuator; SPC actuator (D) The; (D photograph) The photograph of the ofactuation the actuation unit; (unit;E) The (E) input The input voltage voltage given given to the to solenoid the solenoi tod control to control the flow.the flow.

The blocking force and free strain (the displacement) of a single SPC actuator at different loading conditions was performed as shown in Figure4C. Dead weights were attached to a pulley and connected to the tip of the actuator. One end of the actuator was clamped, and the other end was attached to calibrated weights using an inelastic wire. The wire was passed over a pulley to provide a smooth movement. The deflection of the tip of the actuator was recorded corresponding to the original position. The weights attached were from 10 g to 50 g in an increment of 10 g. A preliminary test was done at a pressure of 7 psi and we obtained a deflection of 40, 30, 10, 8 and 7 mm corresponding to the Robotics 2019, 8, 56 8 of 20 loads (10, 20, 30, 40 and 50 g). Most of the experiments were done by providing a pulse actuation, supplying the signal shown in Figure4E to the solenoid and controlling the pressure. The swimming motion of the prototype robot was done as follows: The prototype was actuated by supplying air through an inflation and deflation system. The system was controlled using Arduino Uno board. The pulse width of actuation of the solenoid was varied to see the effect in swimming. The experiment was also conducted by sweeping the required supply pressure that will give the maximum bending and the fastest movement. The actuation time or inflation time (Ti) was 500 ms and the exhaust time or deflation time (Te) was 1200 ms. All the swimming motion tests were performed using a glass fish tank of length 35” (889 mm), width 18” (457 mm) and height 25” (635 mm). The circuit discussed before was used to actuate the jellyfish and determine its properties. The first characterization experiment of the prototype was carried out using a Kinect camera to track the positions. The images were acquired at 12 fps using a camera through MATLAB. A red marker was placed on the FludoJelly to track the position in time. The schematic diagram of the experiment is shown in the figure (Figure4A). A weight of 50 gm was added to the bottom of the prototype to make it neutrally buoyant. Later, a high frame rate camera (Cannon) was used to test the fast motion while varying the supply pressure from 50 psi (345 kPa), 60 psi (414 kPa) and 70 psi (483 kPa). We added another load and tested the robot with 100 g load payload and varying the pressure to see its performance under higher load. One may think that the inflation of the fluidic compartment alone may yield floating of the prototype. However, that is not the case because the jellyfish is carrying a heavy weight (100 g) compared to its own weight (500 g). It will sink to the bottom of the fish tank and it needs to flap the bell to swim vertically. The air supplied to the compartment helps in bending the bell segments, push the water out, and create a vortex. Additionally, the volume change of the chamber will have an effect in lifting the prototype upward (buoyancy effect). Therefore, the vertical swimming of the robot is a contribution due to volume change as well as due to the bending of the actuator. To see the effect of one of them, we secured the prototype at different height and tested the deflection (bending) due to the pressure input. This will be discussed later.

3. Results and Discussion

3.1. Displacement Profile XY Trajectories of the SPC Actuators To understand the behavior of the SPC actuator that is used in the robotic jellyfish, we analyzed the displacement profile using a high speed camera (Phantom Miro eX2). Two circular markers were placed, first at the end of the actuator and second at the mid-point of the actuator. The points tracked through the camera were plotted in MATLAB and shown in Figure5A and the trajectories are shown in Figure5B. To compare with the commonly used multi-chambered soft actuator PneuNet, a prototype with 6 small chambers each 22 24 15 mm3 compartment were made and tested as × × shown in Figure5C. Two points were tracked like the previous case. The trajectories obtained are shown in Figure5D. Three points tracking of the actuator is shown in Figure5E and the corresponding trajectories are shown in Figure5F. The two points and three points tracking were done to check the trajectories along different points. It was observed that forward and backward trajectories of each of these points are not the same. This is due to the slight hysteresis in the pneumatic system. More studies on this structure should be done in the future to see the effect of geometry and materials. Trajectories of the bell segments are one of the important aspects in the motion of robotic jellyfish [20,26,37], and so we added the trajectories of the current actuators. In this work, we used a single chamber instead of a multi-chamber arrangement because unequal expansion of the chambers and delay in inflation of the successive chambers resulted in an undesired motion. Robotics 2019, 8, x FOR PEER REVIEW 9 of 20

Roboticsmulti2019-chamber, 8, 56 arrangement because unequal expansion of the chambers and delay in inflation of the9 of 20 successive chambers resulted in an undesired motion.

(A) (B)

(C) (D)

(E) (F)

FigureFigure 5. 5. Actuators and and their their XY XY trajectories: trajectories: (A,B ()A Actuator,B) Actuator 1 with 1 single with singlechamber chamber and spring and steel, spring steel,(C,D (C) ,ActuatorD) Actuator 2 with 2 with multiple multiple chambers chambers and and two two point point tracked, tracked, (E,F ()E ,ActuatorF) Actuator 2 with 2 with multiple multiple chamberschambers when when three three points points areare trackedtracked along the the edge. edge. 3.2. Position Tracking During Swimming of the Fludojelly

The swimming motion of the FludoJelly was analyzed from recorded videos using Phantom Miro, camera at frame rate 50–1200 fps from experimental setup shown earlier. The videos were analyzed and plotted in MATLAB. The result is shown in Figure6A. FludoJelly was tested at three di fferent Robotics 2019, 8, x FOR PEER REVIEW 10 of 20

3.2. Position Tracking During Swimming of the Fludojelly The swimming motion of the FludoJelly was analyzed from recorded videos using Phantom Miro, Roboticscamera2019 at frame, 8, 56 rate 50–1200 fps from experimental setup shown earlier. The videos were analyzed10 ofand 20 plotted in MATLAB. The result is shown in Figure 6A. FludoJelly was tested at three different pressures pressures50, 60 and 50,70 psi. 60 andThe frequency 70 psi. The of frequencyactuation was of actuation 0.83 Hz, actuation was 0.83 time Hz, of actuation 0.25 s and time pacification of 0.25 s time and pacification1.2 s for all the time experiments. 1.2 s for all theThis experiments. signal shown This in Fig signalure 4E shown was inprovided Figure4 toE wasthe solenoid provided keeping to the Ti = 250 ms for contraction pulse and Te = 1200 ms for relaxation pulse. The result of the swimming test solenoid keeping Ti = 250 ms for contraction pulse and Te = 1200 ms for relaxation pulse. The result of thein underwater swimming environment test in underwater is shown environment in Figure is6A shown. From inthe Figure figure,6A. we From can thesay figure,that the we prototype can say thatshowed the high prototype slope in showed vertical high motion slope (160 in mm/s) vertical with motion respect (160 to time mm /ats) higher with respect pressure to 70 time psi. atIt moved higher pressure500 mm from 70 psi. the It initial moved position 500 mm 100 from mm the (400 initial mm position displacement) 100 mm within (400 mm 2.5 s. displacement) We observed within the motion 2.5 s. Weof the observed prototype the motion while ofcarrying the prototype a payload while mounted carrying at a payload the bottom mounted (Video at S the2 FludoJelly2 bottom (Video 100S2 g FludoJelly2payload.mp4 100 supplementary). g payload.mp4 This supplementary). was consistent with This another was consistent swimming with test another when the swimming weight of testthe whenpayload the was weight 50 g. of The the result payload for was 50 g 50 load g. Theis not result presented, for 50 g but load similar is not increase presented, in butslope similar was obtained increase inas slopethe pressure was obtained was increased. as the pressure The arrows was increased.in the figure The show arrows the in initial the position figure show of the the robot initial during position the ofvertical the robot motion. during In theFig verticalure 6B, motion.the velocity In Figure of the6B, the robot velocity tracked of the from robot an tracked open source from an tracking open sourcesoftware tracking [38] is softwareshown. It [ 38was] is observed shown. It that was as observed the pressure that as was the increased pressure wasfrom increased 50 psi to from 70 psi, 50 the psi tonumber 70 psi, of the cycles number needed of cycles to reach needed the to top reach of the the tank top ofdecreased. the tank Thedecreased. scattered The points scattered are due points to aretracking due to error tracking of the error video of theand videocan be and ignored. can be However, ignored.However, the velocity the harmonics velocity harmonics for each pressure for each pressuretest can be test seen can in be the seen figure in the clearly. figureclearly. Photographs Photographs of the ofrobot the robotat different at diff erenttime timeintervals intervals are also are alsoshown shown in Fig inure Figure 7. We7. We did did not not study study the the vehicle vehicle from from fluid fluid dynamic dynamic point point of of view such asas addedadded mass e effectffect as as explained explained in in[39 [3],9 but], but such such theoretical theoretical framework framework will be will essential be essential to study to such study systems such andsystems this isand left this to futureis left to works. future works.

D (A)

(B)

Figure 6.6. Tracking thethe positionposition ( A(A)) and and velocity velocity (B (B) of) of the the prototype prototype FludoJelly FludoJelly at diat ffdifferenterent pressure pressure 50, 6050, and 60 and 70 psi 70 whilepsi while carrying carrying a dead a dead weight weight of 100 of 100 gram. gram.

Robotics 2019, 8, 56 11 of 20 Robotics 2019, 8, x FOR PEER REVIEW 11 of 20 Robotics 2019, 8, x FOR PEER REVIEW 11 of 20

Figure 7. Photograph of the robot while swimming carrying 100 g load at different supply pressure FigureFigure 7. 7.Photograph Photograph of of the the robot robot whilewhile swimmingswimming carrying carrying 100 100 g gload load at at different different supply supply pressure pressure (50(A), 60(B) and 70 psi(C)) using fast camera for motion capture. All the photos were taken at 1 s (50((50A(),A 60(), 60B()B and) and 70 70 psi( psiC(C)))) using using fast fast cameracamera forfor motion capture. capture. All All th thee photos photos were were taken taken at at1 s 1 s intervals after it start moving. intervalsinterval afters after it it start start moving. moving. The trajectory of the robot is not a perfect vertical motion (z-direction). The robot moves in the TheThe trajectory trajectory of of the the robot robot is is not not a perfecta perfect vertical vertical motion motion (z-direction). (z-direction). The The robot robot moves moves inin the the 3D 3D space slightly. As can be seen in Figure 8, it moves in the x and y directions. Some bell segments space3D space slightly. slightly. As can As be can seen be inseen Figure in Fig8,ure it moves 8, it moves in the in x andthe x y and directions. y direction Somes. Some bell segments bell segments inflate inflate more than others, and this creates some side movement of the robot. The variation is due do moreinflate than more others, than and others this, and creates this somecreates side some movement side movement of the of robot. the robot. The variationThe variation is due is due do slightdo slight differences in the thicknesses of the walls of the air compartments, which indicates that this diffslighterences difference in the thicknessess in the thickness of thees walls of the of wall thes air of compartments,the air compartments, which indicateswhich indicates that this that variable this variable should be carefully designed. The results in Figure 8 are the motion trajectories obtained shouldvariable be carefullyshould be designed. carefully designed. The results The in re Figuresults 8in are Fig theure motion8 are the trajectories motion trajectories obtained obtained from high from high speed camera recorded at 600 fps and plotted with Phantom Camera Control (PCC) speedfrom camera high speed recorded camera at 600 recorded fps and plotted at 600 fps with and Phantom plotted Camera with Phantom Control (PCC)Camera software. Control Our (PCC) high software. Our high speed camera, Phantom Miro can capture a video from 50–1200 fps, which can be speedsoftware. camera, Our Phantom high speed Miro camera, can capture Phantom a videoMiro can from capture 50–1200 a video fps, whichfrom 5 can0–1200 be selectedfps, which depending can be selected depending on the need. onselected the need. depending on the need.

(A) (B) (C) (A) (B) (C) Figure 8. Trajectory of FludoJelly recorded with speed camera and varying input pressure. (A) 50 Psi; FigureFigure 8. 8.Trajectory Trajectory of of FludoJelly FludoJelly recordedrecorded withwith speed camera camera and and varying varying input input pressure. pressure. (A ()A 50) 50Psi; Psi; (B) 60 Psi; (C) 70 Psi. (B)(B 60) 60 Psi; Psi; (C ()C 70) 70 Psi. Psi.

Robotics 2019, 8, 56 12 of 20

Robotics 2019, 8, x FOR PEER REVIEW 12 of 20 3.3. Cyclic Actuation of the FludoJelly 3.3. Cyclic Actuation of the FludoJelly To check the cyclic motion of the robot, tests were done by supplying constant pressure (50, 60 andTo 70check psi) the with cyclic 100 motion g payload of the and robot, controlling tests were duration. done by supplying The pulse constant period waspressure 20% (50, duty 60 cycle and 70at 0.8psi)Hz with frequency 100 g payload (250 ms and on controlling and 1000 msduration. off). The The experiment pulse period was was done 20% as duty follows. cycle Theat 0.8 pulse Hz frequencypressure was (250 given ms on until and the1000 prototype ms off). The reached experiment to the topwas ofdone the as fish follows. tank. OnceThe pulse it reached pressure the was top, giventhe supply until the pressure prototype was reached stopped to allowing the top of the the prototypefish tank. Once to sink. it reached When the it reached top, the tosupply the bottom pressure of wasthe tank,stopped another allowing cycle the of theprototype pulse was to sink. provided, When andit reached this process to the wasbottom done of threethe tank, times. another The results cycle offor the the pulse three was diff provided,erent cyclic and pressures this process obtained was done from three the tracking times. The software results are for shownthe three in different Figure9. cyclicVideo pressures of the test obtained is also provided from the as tracking supplementary software information are shown (Video in Fig S3ure Cyclic 9. Video Swimming of the test FludoJelly is also provided60 psi 1X.mp4). as supplementary As can be seen information in the figure, (Video the S3 e ffCyclicective Swimming height of 400 FludoJelly mm was 60 obtained psi 1X.mp4). for all As three can becycles seen and in the the figure, frequency the effective of swimming height upof 400 and mm sinking was obtained down (one for complete all three cycles cycle) and was the 0.05 frequency Hz. The ofrising swimming phase (orangeup and sinking shaded down region (one in Figurecomplete9) iscycle) the poweredwas 0.05 Hz. cycle The due rising to the phase pulse (orange air pressure shaded regionand the in downward Figure 9) is (blue the powered region) iscycle the sinkingdue to the phase pulse due air to pressure the attached and the weight, downward thereby (blue making region) the issinking-down the sinking phase cycle due longer. to the attached weight, thereby making the sinking-down cycle longer.

FigureFigure 9. CyclicCyclic actuation actuation of of FludoJelly FludoJelly moving moving vertically vertically and and down down in in a a fish fish tank tank when when different different pressure is supplied at 50, 60 and 70 psi and carrying 100 gm mass payload. 3.4. Testing Deflection of the Bell at Different Height in A Water Column 3.4. Testing Deflection of the Bell at Different Height in A Water Column In this test, the FludoJelly was fixed at a specific height inside the fish tank and the deflection of In this test, the FludoJelly was fixed at a specific height inside the fish tank and the deflection of the bell was measured using the camera. The purpose of this test is to investigate whether the water the bell was measured using the camera. The purpose of this test is to investigate whether the water height influenced the bell deformation and quantify the magnitude. The position of the actuator of the height influenced the bell deformation and quantify the magnitude. The position of the actuator of the FludoJelly is like a cantilever beam as shown in Figure 10A. The prototype was clamped as shown in FludoJelly is like a cantilever beam as shown in Figure 10A. The prototype was clamped as shown in Figure 10B and kept inside the fish tank at a different height. The experimental setup consists of a Figure 10B and kept inside the fish tank at a different height. The experimental setup consists of a fish fish tank, the prototype, fixture, solenoid valves and actuation circuit. The actuation circuit consists tank, the prototype, fixture, solenoid valves and actuation circuit. The actuation circuit consists of of Arduino Uno board, pressure sensors and the air compressor as seen in Figure 10C. The actuation Arduino Uno board, pressure sensors and the air compressor as seen in Figure 10C. The actuation sequence was programmed and sent to the Arduino Uno board. The FludoJelly was then actuated at sequence was programmed and sent to the Arduino Uno board. The FludoJelly was then actuated at heights of 150 mm, 300 mm and 450 mm, measured from the bottom of the fish tank by varying the heights of 150 mm, 300 mm and 450 mm, measured from the bottom of the fish tank by varying the gauge pressure at 50, 60 and 70 psi at each height. The deflections of the bell were recorded using a gauge pressure at 50, 60 and 70 psi at each height. The deflections of the bell were recorded using a camera operated at 30 fps. Subsequently, an open source image processing software (Physlets Video camera operated at 30 fps. Subsequently, an open source image processing software (Physlets Video Tracker) was used to track a specific point marked on the FludoJelly and the results were obtained. Tracker) was used to track a specific point marked on the FludoJelly and the results were obtained. Figure 11A represents the deflection vs time characteristics obtained during the actuation of FludoJelly Figure 11A represents the deflection vs time characteristics obtained during the actuation of FludoJelly at 150 mm height from the bottom of the fish tank. Here, we can see that the deflection of the bell at 150 mm height from the bottom of the fish tank. Here, we can see that the deflection of the bell increases as we increase the input air pressure (gauge pressure). A similar trend can be observed from Figure 11B at height 300 mm and Figure 11C at height 450 mm. The importance of this figure

Robotics 2019, 8, 56 13 of 20 increases as we increase the input air pressure (gauge pressure). A similar trend can be observed from FigureRobotics 201911B, at8, x height FOR PEER 300 REVIEW mm and Figure 11C at height 450 mm. The importance of this figure (Figure13 of 11 20) is that it shows that the forward actuation is so quick due to the air pressure, whereas the return is slower(Figure because11) is that the it air shows slowly that leaves the forward the air chamber actuation (it is is so not quick a forced due flow).to the Aair percentage pressure, deflectionwhereas the at eachreturn height is slower and eachbecause pressure the air can slowly be seen leaves from the Figure air chamber 12. The (it maximum is not a forced deformation flow). percentageA percentage of thedeflection bell segment at each is height 59% (the and deflection each pressure is 45 can mm be and seen the fr lengthom Figure of the 12 bell. The segment maximum is 76 deformation mm). From thispercentage experiment, of the we bell can segment say that is as 59% we (the increase deflection the height is 45 ofmm the and jellyfish the length position of insidethe bell the segment fish tank, is the76 mm deflection). From this of the experiment, FludoJelly we also can increases say that as slightly. we increase The minimum the height percentageof the jellyfish deflection position of inside 45% wasthe fish obtained tank, the at deflection the height of of the 450 FludoJelly mm and 50also psi increases pressure. slightly. Supplementary The minimum video percentage is provided deflection for the flappingof 45% was characteristics obtained at the at diheightfferent of height450 mm (Video and 50 S4 psi Height pressure. Deflection Supplementary of the Bell video 1.5X.mp4). is provided for the flapping characteristics at different height (Video S4 Height Deflection of the Bell 1.5X.mp4).

(A) (B)

(C)

Figure 10. 10.Experimental Experimental setup setup for testing for testing bell deformation bell deformation at specific at height specific in a height fish tank. in (A a )fish Schematic tank. (diagramA) Schematic of the diagram cantilever of the beam cantilever at various beam heights at various in water; heights (B) in Experimental water; (B) Experimental setup for the setup test and for (theC) Jellyfishtest and ( prototypeC) Jellyfish with prototype stand and with clamp. stand and clamp.

RoboticsRobotics2019, 8, 2019 56 , 8, x FOR PEER REVIEW 14 14of of2020

(A)

(B)

(C)

FigureFigure 11. Deflection 11. Deflection profile profile at di atff erentdifferent height height above above the the base base of of the the fish fish tanktank ((AA)) 150 mm, mm, ( (BB) )300 300 mm mm and (Cand) 450 (C mm.) 450 Themm. lengthThe length of the of bellthe bell is 76 is mm76 mm and and the the deflection deflection distance distance cancan be normalized by by the length.the length.

Robotics 2019, 8, x FOR PEER REVIEW 15 of 20

Additionally, a maximum deflection of 46 mm was obtained at height of 450 mm and 70 psi pressure. Theoretically, the pressure applied on the FludoJelly at different height can be found by the following relation: 푃 = 𝜌𝑔푧 (3) where 푃 is the pressure, 𝜌 is density of the fluid, 𝑔 is acceleration due to gravity and 푧 is height of the fluid. From Figure 10A, we have H* which represents the total height of the fish tank, 635 mm. H represents the total water level height, ~560 mm. The heights where the FludoJelly was clamped in the

fish tank during experiment were ℎ1 , ℎ2 and ℎ3 . Therefore, the pressure at height h1 is 푃1 = 𝜌𝑔(퐻 − ℎ3) which is less as compared to pressure acting at height ℎ1 and ℎ2. Since the quantity (퐻 − ℎ) increases from ℎ3 to ℎ1, the pressure increases as we go down in the fish tank. Hence, due to the increase in pressure, less deflection is seen at the bottom position as compared to the top ones. This was observed in the experimental results in Figure 11. We noticed that the applied pressure (50–70 psi, equivalent to 3.45 × 105–4.83 × 105 Pa) are much higher than the pressure due to the water height, which 3 Roboticsis decreasing2019, 8, 56 as the height increase (left axis of Figure 12B). The maximum pressure is 4 × 10 Pa, which15 of 20 is 100 times less than the applied pressure. Therefore, the height is not the dominant factor in the fish tank experiment.

(A) (B)

FigureFigure 12. 12.( (AA)) PercentagePercentage deflection deflection vs vs pressure pressure of of the the bell bell (length (length 76 76mm) mm) at different at different height heightss in a in a waterwater column, column, (B(B)) thethe pressurepressure due to to water water height height and and the the pressure pressure applied applied through through the thecompressor compressor showingshowing the the 100 100× orderorder of magnitude difference. difference. × 3.5. Additionally,Comparison of Fludojelly a maximum with deflectionOther Jellyfish of-Like 46 mm Prototypes was obtained at height of 450 mm and 70 psi pressure.Compared Theoretically, to the previousthe pressure robotic applied jellyfish on the prototypes, FludoJelly this at design different (FludoJelly) height can has be the found fastest by the followingmotion. This relation: is due to the quick actuation of the bell segment using compressed air that results in both buoyancy as well as flapping of the bell segmentP = ρgz for swimming vertically, thanks to the SPC (3) actuators/architecture. A dedicated study on the contribution of buoyant force and thrust force due to where P is the pressure, ρ is density of the fluid, g is acceleration due to gravity and z is height of the flapping is needed since the inflatable flexible structure contributes for the two effects the fluid. (flapping and buoyancy). The topic is interesting, but it is beyond the scope of this project objective. From Figure 10A, we have H* which represents the total height of the fish tank, 635 mm. H There are dedicated papers related to this topic and we direct readers to References [39,40]. The represents the total water level height, ~560 mm. The heights where the FludoJelly was clamped in the limitation of our proposed actuation system for the robot is that the inflation of the chambers will fish tank during experiment were h , h and h . Therefore, the pressure at height h1 is P = ρg(H h ) change the density of the entire robot,1 2 which 3may not be desired in some applications. Such1 density− 3 which is less as compared to pressure acting at height h and h . Since the quantity (H h) increases change should be compensated for other desired directional1 2 movements. However, for− vertical fromlocomotionh3 to h1 in, the water, pressure the proposed increases system as we is go a great down choice. in the For fish example tank. Hence,, to save due life to of the a drowned increase in pressure,person such less deflection a vertically is seenswimming, at the bottom fast jellyfish position-like as rescue compared robots to might the top be ones. a good This choice. was observed The inproposed the experimental system could results also inbe Figurea mode 11 of. transport We noticed that that can thebe turned applied on pressureand off as (50–70 needed. psi, Regarding equivalent tovertical 3.45 swimming,105–4.83 10 the5 Pa) graph are i muchn Figure higher 13 shows thanthe comparable pressure due prototypes to the water and the height, swimming which is × × decreasing as the height increase (left axis of Figure 12B). The maximum pressure is 4 103 Pa, which × is 100 times less than the applied pressure. Therefore, the height is not the dominant factor in the fish tank experiment.

3.5. Comparison of Fludojelly with Other Jellyfish-Like Prototypes Compared to the previous robotic jellyfish prototypes, this design (FludoJelly) has the fastest motion. This is due to the quick actuation of the bell segment using compressed air that results in both buoyancy as well as flapping of the bell segment for swimming vertically, thanks to the SPC actuators/architecture. A dedicated study on the contribution of buoyant force and thrust force due to the flapping is needed since the inflatable flexible structure contributes for the two effects (flapping and buoyancy). The topic is interesting, but it is beyond the scope of this project objective. There are dedicated papers related to this topic and we direct readers to References [39,40]. The limitation of our proposed actuation system for the robot is that the inflation of the chambers will change the density of the entire robot, which may not be desired in some applications. Such density change should be compensated for other desired directional movements. However, for vertical locomotion in water, the proposed system is a great choice. For example, to save life of a drowned person such a vertically swimming, fast jellyfish-like rescue robots might be a good choice. The proposed system could also be a mode of transport that can be turned on and off as needed. Regarding vertical swimming, the graph in Robotics 2019, 8, 56 16 of 20

FigureRobotics 13 shows2019, 8, x comparableFOR PEER REVIEW prototypes and the swimming performance of the best biomimetic16 of robotic20 jellyfishes demonstrated to date, such as Robojelly [20] and the large-size robotic jellyfish Cryo [22]. Weperformance did not compare of the otherbest biomimetic types of robotic robotic fish jellyfishes such as demonstrated octopus-like to robots, date, such tuna-like as Robojelly robots, [20] etc. and There arethe some large interesting-size robotic works jellyfish in soft Cryo robot [22]. that We operated did not based compare on other pulsed-jetting types of robotic propulsion fish such [41,42 as] that areoctopus related-like to the robots, current tuna work.-like robot Additionally,s, etc. There are a review some interesting paper on works aquatic in robotssoft robot are that great operated resources based on pulsed-jetting propulsion [41,42] that are related to the current work. Additionally, a review for referring the state-of-the-art marine robots [43]. We only compared jellyfish-like robots where paper on aquatic robots are great resources for referring the state-of-the-art marine robots[43]. We only vertical swimming data are available. This was done to make the paper focused on jellyfish-like robots compared jellyfish-like robots where vertical swimming data are available. This was done to make the and their swimming characteristics [18,20,44–47]. These robots have already shown the potential of paper focused on jellyfish-like robots and their swimming characteristics [18,20,44–47]. These robots usinghave them already for theshown designs the potential of underwater of using vehicles.them for the The designs current of prototype underwater adds vehicles. a new The knowledge current of theprototype robotic jellyfish adds a new using knowledge custom made of the pneumaticrobotic jellyfish soft using actuators custom and made fills pneumatic the gaps betweensoft actuators different designand approaches.fills the gaps Thebetween experiments different fordesign the FludoJellyapproaches. were The allexperiments performed for in the a smallFludoJelly fish tank were that all had a maximumperformed height in a small of 635 fish mm, tank butthat futurehad a maximum test will beheight performed of 635 mm, in abut taller future tank. test Moreover, will be performed testing of a prototypein a taller will tank. be performedMoreover, testing with allof a the prototype electronics will andbe performed pressure with source all onthe boardelectronics as one and self-contained pressure module,source as on with board a previousas one self demonstration-contained module in, aas robotic with a previous fish [35]. demonstration in a robotic fish [35].

(A)

(B)

FigureFigure 13. 13.Comparison Comparison of of swimming swimming jellyfish-likejellyfish-like biomimetic biomimetic soft soft robots. robots. In the In thetop topgraph graph (A) we (A ) we havehave compared compared with with other other prototypes prototypes inin the literature. In In the the bottom bottom graph graph (B) (weB) wehave have excluded excluded CryoCryo (2013) (2013) to to obtain obtain a zoomeda zoomed version version of the graph. All All dimensions dimensions are are in mm in mm and and the thearrow arrow lines lines depictdepict the the size size of theof the robots. robots. (Note: (Note: vertical vertical movement movementvs vs timetime datadata ofof somesome jellyfish-likejellyfish-like robots robots are are not not available or missing in publications, and therefore are not plotted here). available or missing in publications, and therefore are not plotted here).

Robotics 2019, 8, 56 17 of 20

Robotics 2019, 8, x FOR PEER REVIEW 17 of 20 3.6. Peripheral Components and Future Works 3.6. Peripheral Components and Future Works We showed the necessary material, geometry and performance relationship of a small robotic jellyfishWe concentrating showed the on necessary actuation material, under load. geometry The jellyfish and performance inspired robotic relationship system of uses a small a compressed robotic airjellyfish source concentrating for actuation on and actuation needs provisionunder load. for The actuation jellyfish inspired source. robotic Therefore, system we uses tried a tocompressed incorporate theair whole source actuation for actuation circuit and and needs powering provision source for actuation into the source. jellyfish Therefore, model. After we tried a few to incorporate design iterations, the consideringwhole actuatio then size circuit of ourand robotpowering (Figure source 14A), into we the designed jellyfish model. and 3-D After printed a few the design outer iterations, casing for theconsidering model to the protect size of the our electronic robot (Fig circuiture 14A and), we other designed peripheral and 3-D components. printed the outer The diametercasing for ofthe the self-containedmodel to protect structure the electronic is 140 mm circuit in diameter and other mm and peripheral 190 mm compone in height.nts. Its The total diameter weight of is 1.7 the kg includingself-contained all the structure electronics, is 140 battery, mm in solenoid,diameter mm canister and 190 and mm the in silicone height. bell.Its total One weight key componentis 1.7 kg including all the electronics, battery, solenoid, canister and the silicone bell. One key component used used here is a 12 g CO2 cylinder required to actuate the jellyfish untethered (Figure 14B). The here is a 12 g CO2 cylinder required to actuate the jellyfish untethered (Figure 14B). The peripheral peripheral components include rechargeable battery, valves, solenoids, voltage regulator, XBEE components include rechargeable battery, valves, solenoids, voltage regulator, XBEE wireless wireless communication module. However, the leakage of compressed gas and the limited gas amount communication module. However, the leakage of compressed gas and the limited gas amount in the in the CO2 cylinder allows the prototype to operate for only a few cycles. The current size of the CO2 cylinder allows the prototype to operate for only a few cycles. The current size of the onboard onboard actuation unit is shown in Figure 14C. Further research and development will be directed actuation unit is shown in Figure 14C. Further research and development will be directed towards towardsthe peripheral the peripheral components, components, design (especially design (especiallyfocusing on focusingsize and weight), on size directional and weight), movements, directional movements,closed loop closed control, loop and control, to perform anding to fie performingld tests. field tests.

(A) (B) (C)

FigureFigure 14. 14.Self-contained Self-contained structure structure for for FludoJelly: FludoJelly: (A ) (A CAD) CAD design design of the of self-contained the self-contained design, design, (B) the (B) the prototype with circuits, battery, CO2 cylinder and electronics housed in the main body and prototype with circuits, battery, CO2 cylinder and electronics housed in the main body and (C) test in water(C) t ofest thein water self-contained of the self- jellyfish.contained jellyfish.

4.4. Conclusions Conclusion InflatableInflatable structures structures found found in in naturenature werewere reviewed in in this this work. work. These These structures structures provide provide some some advantagesadvantages for for animals animals to to survive survive oror adaptadapt in terrestrial terrestrial or or aquatic aquatic environment environments.s. Here, Here, we we developed developed elastomericelastomeric actuators actuators with with spring spring steels steels (soft pneumatic(soft pneumatic composite, composite, SPC) and SPC) examine and theexamine performance, the e.g.,performance trajectory, and e.g., bendingtrajectory characteristics, and bending characteristicsof the structure., of The the structure.inflation of The the inflation actuators of causes the bendingactuators of causes the structure bending and of the enables structure flapping and enables of the bellflapping segment of the of bell the segment jellyfish-like of the robot,jellyfish pushing-like therobot, surrounding pushing waterthe sur outrounding and creating water aout vortex. and creating The jellyfish-like a vortex. robotThe jellyfish FludoJelly-like showedrobot FludoJelly impressive showed impressive actuation characteristics as well as swimming performance. The prototype has actuation characteristics as well as swimming performance. The prototype has 220 mm in diameter, 220 mm in diameter, was able to carry a 100 g load and swim vertically at a speed of 160 mm/s when was able to carry a 100 g load and swim vertically at a speed of 160 mm/s when tested in a fish tank. tested in a fish tank. This prototype has the fastest motion in ascending vertically compared to the This prototype has the fastest motion in ascending vertically compared to the previous robotic jellyfish. previous robotic jellyfish. We tested the prototype at 50, 60 and 70 psi pressure magnitudes. In all We tested the prototype at 50, 60 and 70 psi pressure magnitudes. In all cases, it showed less time cases, it showed less time for ascending as the input pressure was increased during the test in a fish fortank. ascending Cyclic testing as the of input the prototype pressure (for was ascending increased and during descending) the test inside in a fish a fish tank. tank Cyclicwas also testing done of thefor prototype three cycles (for while ascending carrying and a 100 descending) g load. The inside ascending a fish motion tank was was also faster done due for to three the flapping cycles while of carryingthe bell asegments 100 g load. as well The as ascending buoyancy motiondue to volume was faster change, due whereas to the flappingthe descending of the motion bell segments was asrelatively well as buoyancy slower as dueit was to sinking volume only change, due to whereas gravity. the To descendingcheck the flapping motion characteristics was relatively ofslower the asbell it was segment, sinking we only secured due the to middle gravity. section To check of the the FludoJelly flapping and characteristics submerged it of at the specific bell segment,heights in we securedthe water the middle tank. We section observed of the that FludoJelly as we increased and submerged the height, it at the specific deflection heights increased, in the water and we tank.

Robotics 2019, 8, 56 18 of 20

We observed that as we increased the height, the deflection increased, and we obtained maximum normalized deformation of 59%. Slight variation in geometry such as thickness and air trapped in the elastomer could create unequal expansion of the chambers/actuators (Video S5 High Speed Camera 50 Psi Pressure 5x Slow Playback. mp4, supplementary). This was observed in some of the experiments during testing, which can be avoided by creating distinct air passages with increased thickness of the actuator wall during the construction of the SPC. A compressed gas source that lasts for a longer period of swimming will be addressed in our future work. Currently, the robot is operated while tethered. In order to change direction and swim in 3D, we will need to include additional pumps or valves to control different segments of the bell, which is left for future work. We determined the basic characteristics of the structure qualitatively and quantitatively. This was extremely useful in order for the next version of the robot to be fully autonomous and operated by onboard compressed gas. In addition, sensor integration and control will be the subjects of future research. Jellyfish species such as Aurelia aurita is one of the most efficient swimmers on the planet [47] and more work should be done to realize an underwater vehicle inspired by such species.

Supplementary Materials: The following are available online at http://www.mdpi.com/2218-6581/8/3/56/s1, Video S1 FludoJelly1, SPC actuators.mp4; Video S2 FludoJelly2 100 g payload.mp4; Video S3 Cyclic Swimming FludoJelly 60 psi 1X.mp4; Video S4 Height Deflection of the Bell 1.5X.mp4; Video S5 High Speed Camera 50 Psi Pressure 5x Slow Playback. mp4. Author Contributions: Conceptualization, A.K. and Y.T.; Data curation, A.J.; Formal analysis, Y.T.; Funding acquisition, Y.T.; Investigation, A.J., A.K. and Y.T.; Methodology, A.J., A.K. and Y.T.; Project administration, Y.T.; Resources, Y.T.; Software, A.J. and A.K.; Supervision, Y.T.; Visualization, A.K.; Writing – original draft, A.K.; Writing – review & editing, A.J. and Y.T. Funding: This work is supported by the Office of Naval Research (ONR), Young Investigator Program, under the grant number N00014-15-1-2503. Acknowledgments: We would like to thank McKenna, program manager at ONR Biorobotics program. We would like to thank Shameek Kulkarni for designing CAD design and of the peripheral components. Conflicts of Interest: The authors declare no conflict of interest.

References

1. Bar-Cohen, Y. Biomimetics—using nature to inspire human innovation. Bioinspir. Biomim. 2006, 1, P1–P12. [CrossRef][PubMed] 2. Jusufi, A.; Vogt, D.M.; Wood, R.J.; Lauder, G.V. Undulatory swimming performance and body stiffness modulation in a soft robotic fish-inspired physical model. Soft Robot. 2017, 4, 1–9. [CrossRef][PubMed] 3. Kim, H.-S.; Lee, J.-Y.; Chu, W.-S.; Ahn, S.-H. Design and Fabrication of Soft Morphing Ray Propulsor: Undulator and Oscillator. Soft Robot. 2017, 4, 49–60. [CrossRef][PubMed] 4. Lin, H.-T.; Leisk, G.G.; Trimmer, B. GoQBot: A caterpillar-inspired soft-bodied rolling robot. Bioinspir. Biomim. 2011, 6, 026007. [CrossRef][PubMed] 5. Rus, D.; Tolley, M.T. Design, fabrication and control of soft robots. Nature 2015, 521, 467–475. [CrossRef] [PubMed] 6. Shepherd, R.F.; Stokes, A.A.; Nunes, R.; Whitesides, G.M. Soft machines that are resistant to puncture and that self seal. Adv. Mater. 2013, 25, 6709–6713. [CrossRef][PubMed] 7. Tolley, M.T.; Shepherd, R.F.; Mosadegh, B.; Galloway, K.C.; Wehner, M.; Karpelson, M.; Wood, R.J.; Whitesides, G.M. A resilient, untethered soft robot. Soft Robot. 2014, 1, 213–223. [CrossRef] 8. Mosadegh, B.; Polygerinos, P.; Keplinger, C.; Wennstedt, S.; Shepherd, R.F.; Gupta, U.; Shim, J.; Bertoldi, K.; Walsh, C.J.; Whitesides, G.M. Pneumatic networks for soft robotics that actuate rapidly. Adv. Funct. Mater. 2014, 24, 2163–2170. [CrossRef] 9. Sun, Y.; Song, Y.S.; Paik, J. Characterization of silicone rubber based soft pneumatic actuators. In Proceedings of the 2013 IEEE/RSJ International Conference on Intelligent Robots and Systems, Tokyo, Japan, 3–7 November 2013; pp. 4446–4453. Robotics 2019, 8, 56 19 of 20

10. Lu, X.; Xu, W.; Li, X. Concepts and simulations of a soft robot mimicking human tongue. In Proceedings of the 2015 6th International Conference on Automation, Robotics and Applications (ICARA), Queenstown, New Zealand, 17–19 February 2015; pp. 332–336. 11. Marchese, A.D.; Katzschmann, R.K.; Rus, D. A recipe for soft fluidic elastomer robots. Soft Robot. 2015, 2, 7–25. [CrossRef][PubMed] 12. Shepherd, R.F.; Ilievski, F.; Choi, W.; Morin, S.A.; Stokes, A.A.; Mazzeo, A.D.; Chen, X.; Wang, M.; Whitesides, G.M. Multigait soft robot. Proc. Natl. Acad. Sci. USA 2011, 108, 20400–20403. [CrossRef] [PubMed] 13. Wehner, M.; Truby, R.L.; Fitzgerald, D.J.; Mosadegh, B.; Whitesides, G.M.; Lewis, J.; Wood, R.J. An integrated design and fabrication strategy for entirely soft, autonomous robots. Nature 2016, 536, 451–455. [CrossRef] [PubMed] 14. Katzschmann, R.K.; Marchese, A.D.; Rus, D. Hydraulic autonomous soft robotic fish for 3D swimming. In Experimental Robotics; Springer: Berlin, Germany, 2016; Volume 109, pp. 405–420. 15. Park, Y.-L.; Chen, B.-R.; Pérez-Arancibia, N.O.; Young, D.; Stirling, L.; Wood, R.J.; Goldfield, E.C.; Nagpal, R. Design and control of a bio-inspired soft wearable robotic device for ankle-foot rehabilitation. Bioinspir. Biomim. 2014, 9, 016007. [CrossRef][PubMed] 16. Colin, S.P.; Costello, J.H. Morphology, swimming performance and propulsive mode of six co-occurring hydromedusae. J. Exp. Biol. 2002, 205, 427–437. [PubMed] 17. AquaJellies 2.0 Autonomous behaviour in a collective. 2017. Available online: https://www.festo.com/group/ en/cms/10227.htm (accessed on 30 July 2017). 18. Xiao, J.; Duan, J.; Yu, J. Design and implementation of a novel biomimetic robotic jellyfish. In Proceedings of the IEEE International Conference on Robotics and Biomimetics (ROBIO), Shenzhen, China, 12–14 December 2013; pp. 988–993. 19. Villanueva, A.; Bresser, S.; Chung, S.; Tadesse, Y.; Priya, S. Jellyfish inspired unmanned underwater vehicle. In Electroactive Polymer Actuators and Devices (EAPAD) 2009, Proceedings of the SPIE Smart Structures and Materials + Nondestructive Evaluation and Health Monitoring, San Diego, CA, USA, 8–12 March 2009; SPIE: Bellingham, WA, USA, 2009; Volume 7287, p. 7287. 20. Villanueva, A.; Smith, C.; Priya, S. A biomimetic robotic jellyfish (Robojelly) actuated by shape memory alloy composite actuators. Bioinspir. Biomim. 2011, 6, 036004. [CrossRef][PubMed] 21. Tadesse, Y.; Villanueva, A.; Haines, C.; Novitski, D.; Baughman, R.; Priya, S. Hydrogen-fuel-powered bell segments of biomimetic jellyfish. Smart Mater. Struct. 2012, 21, 045013. [CrossRef] 22. Villanueva, A.A.; Marut, K.J.; Michael, T.; Priya, S. Biomimetic inspired by the Cyanea capillata (Cyro). Bioinspir. Biomim. 2013, 8, 046005. [CrossRef][PubMed] 23. Yang, Y.; Ye, X.; Guo, S. A new type of jellyfish-like microrobot. In Proceedings of the 2007 IEEE International Conference on Integration Technology, Shenzhen, China, 20–24 March 2007; pp. 673–678. 24. Nawroth, J.C.; Lee, H.; Feinberg, A.W.; Ripplinger, C.M.; McCain, M.L.; Grosberg, A.; Dabiri, J.O.; Parker, K.K. A tissue-engineered jellyfish with biomimetic propulsion. Nat. Biotechnol. 2012, 30, 792–797. [CrossRef] [PubMed] 25. Frame, J.; Lopez, N.; Curet, O.; Engeberg, E.D. Thrust force characterization of free-swimming soft robotic jellyfish. Bioinspir. Biomim. 2018, 13, 064001. [CrossRef] 26. Smith, C.; Villanueva, A.; Joshi, K.; Tadesse, Y.; Priya, S. Working principle of bio-inspired shape memory alloy composite actuators. Smart Mater. Struct. 2010, 20, 012001. [CrossRef] 27. Tadesse, Y.; Brennan, J.; Smith, C.; Long, T.E.; Priya, S. Synthesis and characterization of polypyrrole composite actuator for jellyfish unmanned undersea vehicle. In Electroactive Polymer Actuators and Devices (EAPAD) 2009, Proceedings of the SPIE Smart Structures and Materials + Nondestructive Evaluation and Health Monitoring, San Diego, CA, USA, 7–11 March 2010; SPIE: Bellingham, WA, USA, 2010; Volume 7642, p. 7642. 28. Yeom, S.-W.; Oh, I.-K. A biomimetic jellyfish robot based on ionic polymer metal composite actuators. Smart Mater. Struct. 2009, 18, 085002. [CrossRef] 29. HARLAN, R. 10 Weird Self-Inflating Animal Species. 2014. Available online: http://listverse.com/2014/01/03/ 10-weird-self-inflating-animal-species/ (accessed on 30 July 2017). 30. Quin Marshall “Get Down In Jellytown!”. Available online: https://quinsprogress.com/2014/02/ (accessed on 5 June 2019). Robotics 2019, 8, 56 20 of 20

31. Australian Red-eyed Tree Frog (Litoria chloris). Available online: https://en.wikipedia.org/wiki/Vocal_sac# /media/File:Litoria_chloris_calling.jpg (accessed on 5 June 2019). 32. Ilievski, F.; Mazzeo, A.D.; Shepherd, R.F.; Chen, X.; Whitesides, G.M. Soft robotics for chemists. Angew. Chem. 2011, 123, 1930–1935. [CrossRef] 33. Almubarak, Y.; Tadesse, Y. Twisted and coiled polymer (TCP) muscles embedded in silicone elastomer for use in soft robot. Int. J. Intell. Robot. Appl. 2017, 1, 352–368. [CrossRef] 34. Onal, C.D.; Chen, X.; Whitesides, G.M.; Rus, D. Soft mobile robots with on-board chemical pressure generation, In Robotics Research; Springer: Berlin, Germany, 2017; Volume 100, pp. 525–540. 35. Marchese, A.D.; Onal, C.D.; Rus, D. Autonomous soft robotic fish capable of escape maneuvers using fluidic elastomer actuators. Soft Robot. 2014, 1, 75–87. [CrossRef][PubMed] 36. Marchese, A.D.; Rus, D. Design, kinematics, and control of a soft spatial fluidic elastomer manipulator. Int. J. Robot. Res. 2016, 35, 840–869. [CrossRef] 37. Villanueva, A.; Vlachos, P.; Priya, S. Flexible margin kinematics and vortex formation of Aurelia aurita and Robojelly. PLoS ONE 2014, 9, e98310. [CrossRef][PubMed] 38. Douglas Brown, Tracker: Open source video analysis and modeling tool, Available online:. Available online: http://physlets.org/tracker/ (accessed on 18 February 2018). 39. Giorgio-Serchi, F.; Weymouth, G.D. Underwater soft robotics, the benefit of body-shape variations in aquatic propulsion. In Soft Robotics: Trends, Applications and Challenges; Springer: Berlin, Germany, 2017; Volume 17, pp. 37–46. 40. Renda, F.; Serchi, F.G.; Boyer, F.; Laschi, C. Structural dynamics of a pulsed-jet propulsion system for underwater soft robots. Int. J. Adv. Robot. Syst. 2015, 12, 68. [CrossRef] 41. Giorgio-Serchi, F.; Lidtke, A.K.; Weymouth, G.D. A soft aquatic actuator for unsteady peak power amplification. IEEE/ASME Trans. Mechatron. 2018, 23, 2968–2973. [CrossRef] 42. Keithly, D.; Whitehead, J.; Voinea, A.; Horna, D.; Hollenberg, S.; Peck, M.; Pikul, J.; Shepherd, R.F. A cephalopod-inspired combustion powered hydro-jet engine using soft actuators. Extreme Mech. Lett. 2018, 20, 1–8. [CrossRef] 43. Salazar, R.; Fuentes, V.; Abdelkefi, A. Classification of biological and bioinspired aquatic systems: A review. Ocean Eng. 2018, 148, 75–114. [CrossRef] 44. Akle, B.; Najem, J.; Leo, D.; Blottman, J. Design and development of bio-inspired underwater jellyfish like robot using ionic polymer metal composite (IPMC) actuators. In Electroactive Polymer Actuators and Devices (EAPAD) 2011, Proceedings of the SPIE Smart Structures and Materials + Nondestructive Evaluation and Health Monitoring, San Diego, CA, USA, 6–10 March, 2011; SPIE: Bellingham, WA, USA, 2011; Volume 7976, p. 7976. 45. Godaba, H.; Li, J.; Wang, Y.; Zhu, J. A soft jellyfish robot driven by a dielectric elastomer actuator. In IEEE Robot. Auto. Lett. 2016, 1, 624–631. [CrossRef] 46. Marut, K.; Stewart, C.; Michael, T.; Villanueva, A.; Priya, S. A jellyfish-inspired jet propulsion robot actuated by an iris mechanism. Smart Mater. Struct. 2013, 22, 094021. [CrossRef] 47. Gemmell, B.J.; Costello, J.H.; Colin, S.P.; Stewart, C.J.; Dabiri, J.O.; Tafti, D.; Priya, S. Passive energy recapture in jellyfish contributes to propulsive advantage over other metazoans. Proc. Natl. Acad. Sci. USA 2013, 110, 17904–17909. [CrossRef][PubMed]

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).