A Wirelessly Controlled Shape-Memory Alloy-Based Bistable Metal Swimming Device

Shape memory Nitinol has long been used for actuation. However, utilizing Nitinol to fabricate novel devices for various applications is a challenge, but has shown incredible promise and impacts. Bistable metal strips are widely adopted for shape morphing purposes (primarily in kid's toys, e.g., snap bracelets) due to their easy and robust transformation between two states. Herein, Nitinol shape memory alloy and bistable metal strip are combined to fabricate a swimming actuator with both slow moving and fast snapping capability, akin to an octopus swimming slowly in water, but quickly moving upon encountering a threat. The actuator developed here can also swim in multiple directions, all controlled by a wireless module. Furthermore, it is demonstrated that an onboard sensor can be incorporated for potential environmental monitoring applications. The fact that the device developed here has no mechanical parts, makes this an interesting potential alternative to more expensive, and energy consuming boats. A preprint version of the article can be found at: https://www.authorea.com/doi/full/10.22541/au.164199106.60208565.


Introduction
For centuries, humans have been fascinated with studying nature and natural processes; this most certainly began out of curiosity, although in many cases led to a desire for mimicry. [1] For example, primitive, early humans would move, dress, and act like animals to improve their hunting successes, while modern day mimicry is focused on the development of new synthetic materials for improving human health and quality of life. [2] Of course, nature has the ability to adapt to changes in environmental conditions and added pressures via many years and generations of evolution, while humans are left primarily using rational thought, science, and engineering for adaptation. A great example of nature adapting to their environment for self-gain is the octopus. The octopus is an intelligent marine creature that has evolved the ability to change color and eject ink when in danger in order to escape and preserve its well-being. Octopuses also exhibit high dexterity, capable of reaching, [3] grabbing, [4] swimming, [5] and walking. [6] Interestingly, they have the ability to swim slowly to disguise themselves as floating algae in ocean currents while maintaining the ability to quickly swim away in a moment by contracting their bodies and whipping their tentacles. Here, we introduce a device composed of Nitinol wire (a shape memory material) and bistable metal strips to generate a device that is capable of mimicking the swimming behavior of an octopus by exhibiting the ability to swim slow and "instantaneously" fast by simply changing how the device's components are electrically stimulated.
Nitinol, generated by alloying nickel and titanium, has many interesting uses and properties, e.g., shape memory, superelasticity, anticorrosion, and biocompatibility. [7] As a result of these interesting properties and attributes, it has long been used for applications in aeronautical and space technologies, automobile industries, medical devices, and civilian products. [8] Of importance to this investigation is Nitinol's shape memory properties, which we propose harnessing for this new approach to actuation. Nitinol can be "trained" to adopt a desired permanent shape when it is fixed while heated above the austenite finish temperature (Af ) and then rapidly cooled below the martensite finish temperature (Mf ). [9] Below Mf, the Nitinol can be reshaped into temporary shapes and upon heating above the austenite start temperature (As), a phase transition from martensite to austenite is initiated, and the Nitinol begins to recover its permanent shape. Importantly, when the temperature is lowered to martensite start temperature (Ms), the Nitinol then can be reshaped again. This shape memory behavior is key to our actuation device. Heating of Nitinol can be achieved by direct heating, or via resistive/ Joule heating by applying a voltage to the wire that generates a subsequent current. For this study, we were interested in using Joule heating to trigger the shape memory behavior from the Nitinol.
Bistable materials are known to exist in two stable states that could differ in conformation dramatically. Very common examples of bistable materials are slap bracelets (a child's toy) and metal measuring tapes. In these examples, an external stimulus (force) can be used to trigger the metal strip to quickly morph from one stable state (e.g., extended state) to the other (e.g., coiled state) in what is often called "snap-through" DOI: 10.1002/aisy.202100251 Shape memory Nitinol has long been used for actuation. However, utilizing Nitinol to fabricate novel devices for various applications is a challenge, but has shown incredible promise and impacts. Bistable metal strips are widely adopted for shape morphing purposes (primarily in kid's toys, e.g., snap bracelets) due to their easy and robust transformation between two states. Herein, Nitinol shape memory alloy and bistable metal strip are combined to fabricate a swimming actuator with both slow moving and fast snapping capability, akin to an octopus swimming slowly in water, but quickly moving upon encountering a threat. The actuator developed here can also swim in multiple directions, all controlled by a wireless module. Furthermore, it is demonstrated that an onboard sensor can be incorporated for potential environmental monitoring applications. The fact that the device developed here has no mechanical parts, makes this an interesting potential alternative to more expensive, and energy consuming boats. A preprint version of the article can be found at: https://www.authorea.com/doi/full/10. 22541/au.164199106.60208565.
behavior. [10] This sudden and powerful actuation is intriguing if it can be harnessed and utilized-in our case for shape changing devices [11] that can be used for swimming.
In this work, we propose harnessing the power of shape memory materials for actuation (and ultimately swimming) by coupling Nitinol wire to bistable materials in a device. By combining these two classes of materials in a device, and utilizing the actuation afforded by Nitinol and the snapping power of bistable metal strips, we developed a novel actuator that mimics the slow and fast swimming behavior of an octopus. Furthermore, by controlling and moving the arms independently (via electrical stimulation), the device can navigate through space in any direction. Additionally, a wireless control system, equipped with a rechargeable battery, was designed to make the device totally autonomous, and untethered from external wires that otherwise would be needed for control. Finally, we demonstrate that we can integrate etalon-based sensing devices [12] into the construct to monitor water pH and ionic strength. In this case, the etalons tethered to the swimming device change color (that can be captured with an onboard camera) as the water properties change. Successful integration of these sensing devices into the construct demonstrates the potential utility of these devices for environmental monitoring applications as well as others.

Results and Discussion
For this study, in order to generate devices capable of exhibiting slow and fast swimming behavior upon electrical stimulation, the power of Nitinol and bistable strips needed to be harnessed using careful device design principles. The design we converged on after many iterations is shown schematically in Figure 1. As can be seen, four pieces of preset short helical Nitinol wires were stretched and attached to the tip of a bistable metal strip. When a voltage was applied to the Nitinol wires, e.g., the front left (FL) piece, the FL Nitinol wire heats up and will shorten, thus pulling the tip of the bistable metal strip to snap and coil (trigger process). This snapping process results in lengthening/straightening of the back left (BL) Nitinol wire. The shortening/ contraction of the BL Nitinol wire can be realized by application of a voltage/heat, thus pulling the tip of the bistable metal strip to uncoil and return to a flat state (reset process). This reset process once again lengthens the FL Nitinol, which prepares the device for its next trigger process. Thus, by carefully (and independently) controlling the conformational state of the four pieces of Nitinol wires with voltage, the state of the bistable strip can be manipulated and controlled allowing slow and fast movement of the device's arms, much like the control an octopus has over its tentacles. However, to understand how all of the pieces of the device work together to achieve the desired behavior, a detailed examination of the device's components is required, as detailed below.

Metal Strip Shape and Trigger/Reset Angle
Initial studies focused on investigating how the shape of the bistable metal strip and trigger/reset angles impacted its triggering and resetting force. Here, "trigger force" refers to the force required to cause the bistable metal strip to rapidly coil, doing work in the process. "Reset force" refers to the force required to bring the bistable metal strip back to its extended state so that it is ready to be triggered again. In terms of "shape," we focused on changing the taper ratios of the bistable metal strips while fixing the total length, as shown in Figure 2. To measure the forces generated from the snapping action of the bistable strip, an inelastic thread attached to the tip of the bistable metal strip was attached to a force detector, and the trigger/reset forces were measured by pulling the thread ( Figure 3). When the thread was pulled, a force-time curve was recorded, as shown in Figure S1, Supporting Information, indicating how forces evolved during the trigger/reset processes. Different pulling rates were tested, and we found pulling rates had no effect on the minimum force required for trigger and reset ( Figure S2, Supporting Information). A pulling rate of %5 cm s À1 was chosen for these experiments due to ease of reproducibility of this rate. The minimum forces required for trigger and reset were plotted against the taper ratio. We concluded from the data in Figure 4 that bistable metal strips with larger taper ratios yielded smaller trigger and reset forces. When the metal strip had a larger taper ratio, we noticed a flatter surface at the tip, which results in easier actuation due to smaller force required to transform the tip from one curvature to the other.
Meanwhile, we investigated how the angle that the force was applied to the bistable metal strip, via the thread, impacted the magnitude of the trigger and reset forces. As shown in Figure 3, using the extended state of the bistable metal strip as a reference, we varied the angle of force application to trigger/reset the bistable metal strips. Three different angles were investigated for both the trigger and reset processes. As can be seen in Figure 5, larger trigger/reset angles led to smaller forces required for triggering and resetting. This could be explained from the force analyses of the process. When the angle of the applied force is large, the effective normal force, which is the projection of the vector force applied onto the tip of the metal, is likewise large. As a result, if the normal force required to trigger or reset the metal strip remains the same, larger angles lead to smaller vector force required. From the result, we concluded that we can minimize the force by maximizing the angle of force application approaching 90 . However, the angle was limited by the actual reasonable dimensions of the device. For our device design, considering the dimensional constraints and aesthetic aspects, we used 45 for triggering and 60 for resetting. www.advancedsciencenews.com www.advintellsyst.com

Energy Output
While the above investigation revealed the impact of the taper ratio of the bistable metal strip on the trigger and rest forces, it also revealed that the taper ratio impacted the energy that can be generated as a result of the snapping process. To further understand the relationship between snapping energy output and taper ratio, we performed energy output tests. It should be noted here that the benefit of using bistable metal strip for device propulsion is its fast and powerful snapping behavior with a relatively small amount of energy input required. Hence, we wanted to retain good snapping performance while minimizing the magnitude of the trigger and reset forces. To accomplish this, we used a regular ping pong ball that was hung over the top side of the bistable metal strip, as shown in Figure S3, Supporting Information. By triggering the tip of the bistable metal strip, the strip snapped, coiled, and struck the center of the ping pong ball, sending it into motion. The ping pong ball then moved in a circular path, like a pendulum, and eventually reached its highest point before swinging back. Assuming no energy loss during the process, the energy output of the bistable metal strip is directly proportional to the maximum height the ping pong ball reached. As shown in Figure 6, the height the ping pong ball can reach (and the energy generated) greatly decreased with larger taper ratios, i.e., larger taper ratios yielded lower snapping energy output. In fact, the 7-0 taper ratio metal strip did not have enough energy to move the ping pong ball. As snapping energy is essential to the performance of the swimming device, we decided to maximize the energy output by adopting a 4-3 taper ratio design for our future devices.

Phase Transition Temperature
We first investigated the phase transition process using differential scanning calorimetry (DSC), as shown in Figure S4, Supporting Information. From the DSC, we observed the phase transition to be %60 C. We also performed a water bath test on the Nitinol ( Figure S5a, Supporting Information). Specifically, the Nitinol was set into a helical structure (contracted state) in a furnace and subsequently stretched into an extended state. Then, the extended Nitinol wire was immersed into a beaker of water for 10 s at various temperatures and taken out for length determination. From the data ( Figure S5b, Supporting Information), we observed a sharp length change at %60 C, which was consistent with the DSC test.

Actuation Force Output at Different Voltages
The forces the Nitinol wire can exert upon temperature-induced contraction/shortening determine its ability to apply force to the bistable metal strip for triggering and resetting. Hence, we investigated the force output of the Nitinol wire when stimulated to contract upon application of a voltage, which induced a  Theoretically, the angle (θ) can range from 0 to 180 . However, restricted by the device design, the angle is below 90 (blue: bistable metal strip; black: pivot fixed onto the middle of the metal strip; silver: pulling thread; box with "F": force sensor).
www.advancedsciencenews.com www.advintellsyst.com temperature change above the phase transition temperature. For these experiments, the Nitinol wire was connected to a force sensor, and was not allowed to contract upon stimulation ( Figure S6a, Supporting Information). Different voltages were applied until the force reached 4.2 N, which was the minimum triggering force for the 4-3 taper ratio bistable metal strip; the  www.advancedsciencenews.com www.advintellsyst.com 4-3 taper ratio was determined to have optimal performance (see above). Once the mentioned force was reached, the applied voltage was removed, and the Nitinol wire was allowed to cool. As can be seen in Figure S6b, Supporting Information, higher voltages applied to the Nitinol decreased the time required for the actuation force of 4.2 N to be reached, which indicated faster actuation. However, the higher applied voltage led to an increase in the time required for the Nitinol to cool after removing the applied voltage. This could be explained considering the Joule heating effect. Assuming that the resistance of Nitinol isn't altered upon heating, according to Ohm's law, the current increases with increasing voltage, producing more heat in a unit time, allowing the Nitinol to reach its phase transition, when the voltage was higher, the current was higher, thus more heat was produced in a unit time, allowing the Nitinol to reach its phase transition temperature relatively fast. However, this excess heat needs to be dissipated from the Nitinol wire to allow for its reversibility ( Figure S6c, Supporting Information). More detailed discussion and calculations are shown in Figure S7, Supporting Information.

Actuation Durability
Durability, such as the number of times the Nitinol can contract/ extend upon stimulation before failure, is an important parameter to study for the devices being generated here. Here, two sets of consecutive tests were performed using the same setup as in the previous actuation force test ( Figure S6a, Supporting Information). The voltage was applied for 10 s and removed to allow the Nitinol wire to cool. From Figure 7a, we observed an %8.3% force decrease within the first four actuations, followed by a relative stabilization of the force decrease, i.e., an additional 14.8% decrease was observed after the subsequent 82 cycles.
Then the same piece of Nitinol was used to determine its ability to reach the requisite 4.2 N force needed to trigger the 4-3 taper ratio bistable metal strip. From Figure 7b, we can see the Nitinol wire was able to reach 4.2 N after 222 repeats with no observable failure. Combining two consecutive tests, the Nitinol wire could at least endure 300 cycles of successful actuation. The exact number of cycles to reach failure was not investigated, but could be easily determined in the future.

Underwater Actuation
To generate a swimming device, it must be able to operate immersed in water at various water conditions, e.g., temperature and currents. One advantage of having water present is the quick cooling of the heated Nitinol compared to Nitinol in air. This is due to the fact that water has 4 times larger heat capacity and 25 times larger heat conductivity than air, which is beneficial for multiple consecutive actuations. However, the drawback is that water brings the Nitinol wire temperature down too quickly when the water temperature is low or when there is turbulence. Hence, we performed some actuation tests by immersing the Nitinol wire under water at room temperature and heating by application of a voltage. The results showed that a much higher voltage was required to achieve similar Nitinol contraction compared to when  www.advancedsciencenews.com www.advintellsyst.com the Nitinol was heated in air. Also, we can clearly see water convection on the surface of the Nitinol wire, indicating quick and large heat loss (Video S1, Supporting Information). With slight turbulence in the water, the Nitinol was not able to actuate at all, even at much higher voltage due to faster heat loss (Video S2, Supporting Information), which made the actuation process unpredictable and uncontrollable. In order to hinder this heat loss process, and to provide the Nitinol wire with a stable, controllable, and predictable actuation environment, we coated the Nitinol wire with a layer of polyacrylamide (PAAm) hydrogel (Figure 8a). A hydrogel was chosen as the coating material because it acts as a physical barrier to hinder the heat loss, but at the same time it can still dissipate the heat because it is a water-rich gel. As shown in Figure 8b,c, with this hydrogel coating, the actuations were successfully observed and were stable even in turbulence while the one without hydrogel coating completely failed to actuate (Video S3-S4, Supporting Information).

Assembly of the Device
The device was fabricated based on the aforementioned test results. The main body of the device was composed of a bistable metal strip with a pivot point going through the strip perpendicularly. Nitinol wires were connected at both tips of the metal strip against the pivot to set the angle required for triggering and resetting. A wireless control system was designed and assembled, as shown in Figure 9a. To increase the propelling ability, polydimethylsiloxane (PDMS) sheets with rigid plastic (3M PP2950 film) strips embedded were fabricated, which were inspired by fish fins. A lithium ion polymer battery was used as a portable power source.

Swimming Demonstration
Swimming behavior was achieved with wireless control and multiple swimming modes (Video S5-S7, Supporting Information). The swimming device was immersed under water with a Styrofoam box floating on water carrying all the electronics (Figure 9b). By controlling the individual electronic switches, slow movement and faster snapping actions could be realized. When both arms were working together, the swimming device could swim forward. When triggered to swim slowly, we determined the swimming speed to be 1.2 cm s À1 (Video S8, Supporting Information), while it was increased to 5.5 cm s À1 upon triggering fast swimming (Video S9, Supporting Information) (calculation see Figure S8, Supporting Information). While with only one arm working, the device could turn. In the slow swimming speed regime, we observed the device could turn 180 within 10 s (Video S10, Supporting Information). It should be noted here that the slow swimming speed was measured after two consecutive stimulation cycles with about 1.4 s between the cycles; each actuation was a result of application of voltage (10% power output of the battery) for 0.3 s. The time interval between stimulation cycles, how long the voltage was applied, and how much percentage of the battery power used for the stimulation can all affect the device swimming speed in the "slow swimming speed" regime. The swimming speed for the device in the "fast swimming www.advancedsciencenews.com www.advintellsyst.com speed" regime was a result of a single stimulation/snapping event. In both regimes, the water temperature can also impact the swimming speed. We point out that all of these parameters can be easily controlled and/or programmed into the electronics of the swimming device. By combining different modes of actuation, the swimming device could easily navigate a water tank by swimming straight, speeding up, changing direction as well as slowing down.

Sensing Application
Microgel-based etalons were constructed by sandwiching poly(Nisopropylacrylamide) (pNIPAm)-based microgels between two thin Au layers. In addition to the native thermoresponsivity of pNIPAm, further responsivity can be imparted to microgels via copolymerization. For example, pNIPAm-based microgels can collapse and swell upon heating and cooling, respectively, while also exhibiting pH-and ionic strength-dependent solvation states by incorporating acrylic acid (AAc) into the pNIPAm microgels. Such responsive microgels in etalons allow their visual color, and peaks in reflectance spectra, to shift upon application of any of these stimuli, allowing the color to be correlated to the composition of the water and its temperature. This is due to the microgel solvation state mediating the distance between the etalon's Au layers, which are responsible for the etalon color; by changing the Au-Au distance, the color of the device changes. We can predict the etalons' optical properties by Equation (1) where λ is the wavelength of light being reflected, n is the refractive index of the microgel layer, d is the distance between the two Au layers (thickness of the microgel layer), θ is the angle of incident light, and m is the order of the reflected peak (e.g., 1, 2, 3,…). As mentioned above, the Au-Au distance (d) can be tuned by exposing the microgel-based etalons to different stimuli. In this work, we focused on detecting pH and ionic strength changes, yielding a shift in the wavelengths of light reflected from the etalon and a concomitant visual color change. [13] 2.4.1. Device Integration A 3D printed holder was made to accommodate etalon devices, a camera and a lighting system. The camera and lighting system were connected to the wireless control module and could take pictures of the etalons to monitor its color change as the water changes pH and ionic strength. The holder was attached to the Styrofoam boat body and immersed in water (Figure 10a).

Salt and pH Sensing
Etalons with salt and pH responsivity pNIPAm-co-10% AAc were prepared following the group's previous fabrication procedure. After we change the solution media from deionized water (DI water) to 0.5 M sodium chloride (NaCl) solution, the color of the etalon chips changed from green to red in 260 s and could return back to green in 260 s when put back to DI water (Figure 10b). Similarly, when the pH was changed from 6.02 to 2.69, the color of the etalon chips changed from green to red in 620 s and could change from red to green in 620 s (Figure 10c). The swimming device served as a platform for different technologies and could potentially be applied in more research areas. www.advancedsciencenews.com www.advintellsyst.com

Conclusions
In summary, a wirelessly controlled, programmable, swimming device was developed. The properties of the Nitinol alloy and the bistable metal strips were investigated to determine the optimal device design. Fish fin-like PDMS flippers were also designed and added to the bistable metal strips to increase the swimming performance. The swimming device could move in all directions, controlled wirelessly and easily programmable due to the untethered design with a portable battery and wireless control module. We also showcased the successful integration with an etalonbased sensing platform, which can potentially inspire more innovative future applications. With this work, we hope to inspire related research that combines the use of various functional materials in a novel design to achieve higher functionality and to access myriad applications. We showed that the swimming device can be used to carry a sensor to an aquatic environment for monitoring applications. While the demonstration here was done in a relatively "pristine" environment, complexity can be introduced into our devices (and systems) to work in "real world" environments. For example, multiple swimming devices carrying sensors can be wirelessly tethered together such that the devices can communicate with one another, and can be sequestered to a specific site if high density monitoring is needed at a particular time. A GPS and various other camera systems (and sensors) can be added to the device as needed, depending on the application. However, we do need to consider the long time memory loss of the Nitinol wire after repeated actuations in an uncontrolled environment, which was not completely investigated in this work. Future research should look into this and come up with an estimated lifetime for the swimming device in various environmental conditions. Finally, a method for recharging the battery needs to be considered. A straightforward solution might be adding solar panels, which can be a future improvement for the device. More importantly, biofouling can be a big challenge for devices that are deployed and immersed in a water environment for long periods of time. Special protective coatings or antifouling design for the device should be investigated.

Experimental Section
Bistable Metal Strip Characterization: Bistable metal strips were purchased from Amazon (Seattle, Washington, USA), originally sold as slap/snap bracelets. We removed the plastic covers to obtain the metal strips with 21 cm in length, 2.5 cm in width, and 0.15 mm in thickness. The metal strip has curvature on both long and short directions. The curvature for the long side is 1.5 and 2.4 cm for the short side. A pair of tin snips was used to cut the metal strips into different shapes with different taper ratios. A small hole was punctured on the tip of the metal strip to allow us to fix a piece of commercially available fishing thread (Beadalon Supplemax JNX0.25W-F, Coatesville, Pennsylvania, USA) with 0.01 in. diameter. We then screw a nail in the middle of the metal strip which acts as a pivot and allow its fixture on the clamp. The other end of the thread was fixed onto a force sensor (Vernier Dual-Range Force Sensor, Beaverton, Oregon, USA). A pulley was used to adjust and fix the force angles. The thread was pulled until the metal strip snaps or resets and the forces were recorded in the Logger Pro software for later analysis.
Nitinol Wire Permanent Shape Setting: Nitinol wire with 0.02 in. diameter was purchased from McMaster-Carr (Elmhurst, Illinois, USA). The Nitinol wire was then coiled onto a 5 mm metal hex wrench with a pitch of 5 mm. The two ends were fixed using copper wire. Then we kept it inside a furnace (Thermolyne FB1315M, Thermo Scientific, Waltham, Massachusetts, USA) preheated to 510 C for 30 min. It was quickly chilled into room temperature (%24 C) tap water until it was cooled.
Nitinol Wire Characterization: Nitinol phase transition temperature was tested both by DSC and by water bath test. Nitinol sample (22.86 mg) was cut from the preset Nitinol coil. Mettler Polymer DSC (Mettler Toledo, Columbus, Ohio, USA) using STARe software (version 16.10) was used to perform DSC test by sweeping the temperature between 0 and 150 C at a heating rate of 20 C min À1 . For the water bath test, we heated www.advancedsciencenews.com www.advintellsyst.com a beaker of water while monitoring the temperature. The coiled Nitinol wire was stretched, immersed it into a water bath at different temperatures, and recorded the length change.
Nitinol actuation voltage and durability tests were done with the same setup. Prestretched Nitinol wire was fixed one end to a monkey bar, the other to a force sensor. Powerstat (Staco Energy Products Co. Type 3PN1010, Miamisburg, Ohio, USA) was used to deliver power. The on/ off operations were performed by hands, which can be inconsistent between repeats. The actual on/off can be interpreted from the force/time graph.
Fabrication of PAAm Hydrogel-Coated Nitinol Wire Coil: All the chemicals were purchased from Sigma-Aldrich (St. Louis, Missouri, USA) and used as received. Milli-Q water was used unless otherwise stated. The PAAm hydrogel-coated Nitinol wire coil was fabricated by wrapping the hydrogel with a thin long channel in the middle onto the coiled-shaped Nitinol wire. First, a presolution was prepared for hydrogel with 0.5331 g of acrylamide (AAm) as monomer, 1.156 mg (0.1 mol% of AAm) of N,N 0 -Methylenebis(acrylamide) (BIS) as cross-linker, and 13.5 mg of potassium persulfate (KPS) as initiator. DI water was added to afford a 2.5 mL solution mixture for later use. This formula was chosen based on many trails. The as-prepared hydrogels were strong so they did not break easily but still have sufficient water content without swelling too much, which would hinder the Nitinol movement. To prepare the long hydrogel with a thin channel inside, we used a commercially available fishing thread (Beadalon Supplemax JNX0.25W-F, Coatesville, Pennsylvania, USA) with 0.01 in. diameter as a template. A plastic tube with a 3.20 mm inner diameter was used. We horizontally clamped the plastic tube, put the fishing thread inside, and fixed the two ends onto monkey bars. Then we pulled the thread to give it a bit tension to make sure it is straight inside the tube. The position of the tube and the thread was carefully adjusted so that the thread is going through the center of the tube ( Figure S9, Supporting Information). A 20 μL N,N,N 0 ,N 0 -tetramethyl ethylenediamine (TEMED) was taken as an accelerator and mixed with the prepared presolution. Some solution was taken using a syringe and slowly injected it into the tube. After polymerization was completed and the hydrogel formed, the thread was slowly pulled to remove the hydrogel out of the tube and then the thread was gently removed inside which should leave a vacant channel inside. Lastly, the Nitinol wire was slowly slid into the hydrogel through the channel.
Fabrication of PDMS Fish Fins: SYLGARD 184 Silicone Elastomer Kit from Dow Inc. (Midland, Michigan, USA) was purchased from Sigma-Aldrich. Polymeric base and curing agent were mixed at 10:1 ratio. Some glitter powder (purchased from Dollarama, Montreal, Québec, Canada) and some crystal violet (Sigma-Aldrich) were added for easy visualization under water. The mixture was mixed well and poured into a large Petri dish. The Petri dish was put on a heating plate overnight at 65 C. A few 3 mm wide plastic strips were cut from a piece of transparency film (3M PP2950, Saint Paul, Minnesota, USA). The strips were placed onto the PDMS and poured into another layer of PDMS presolution, and cured overnight similar to previous steps. The cured PDMS sheet was cut into the fish fin shape. The final fish fins were attached onto the two ends of the bistable metal strip using superglue.
Design and Assembly of the Wireless Swimming Device: A lithium ion polymer battery (FCONEGY 11.1 V 3S 7000 mAh 40 C Lipo RC Rechargeable Battery), a wireless Wi-Fi module with a camera (CANADUINO ESP32-CAM Wi-Fi Bluetooth Module with 2MP Camera), a voltage limiter (KeeYees MP1584EN Mini Step Down Buck Converter Adjustable DC to DC 4.5-28 V to 0.8-20 V Voltage Regulator Module), and four electronic switches (Mosfet Driver Module Dual High-Power 0-20 kHz FET PWM Trigger Switch Driver Module DC 5V-36V 15A) were purchased from Amazon. One outlet from the battery was connected with the voltage limiter and then connected to the wireless Wi-Fi module. The Wi-Fi module was then connected to four electronic switches which are also powered directly by the battery through the second outlet. The electronic switches had wires to connect with the four pieces of Nitinol wires. The Wi-Fi module was programmed using Spyder 4.2.5 on Anaconda-Navigator ( Figure S10, Supporting Information).
3D Printing of the Device Parts: The parts were designed using FreeCAD. Hollow design was adopted to hide the connecting wire in order to avoid cluster and keep the swimming device neat ( Figure S11, Supporting Information). Prusa 3D printer (Prusa Research Company, Prague, Czech Republic) was used for printing using polylactic acid (PLA) as the printing material.
Design and Assembly of the Sensing Platform: The holder was designed using FreeCAD and 3D printed using a Prusa 3D printer. The holder was then attached to the Styrofoam boat and adjusted so that the etalon chips could be fully immersed in water. The camera and LED lighting system were included in the wireless Wi-Fi module and were integrated and programmed into the whole control system so that the picture taking and lighting can work together to get good quality pictures.
Salt and pH Sensing: Etalons were fabricated based on previous procedure ( Figure S12, Supporting Information). For the salt response investigation, the swimming device was put into DI water with the etalons fully immersed under water overnight. Pictures were taken to record the initial color. Then the media was changed to 0.5 M NaCl solution, and pictures taken every minute to record the color change. Similarly, for pH sensing, pH 6.02 was used to record initial color. Then the media was adjusted to pH 2.69 and then 10.99. Color evolution was recorded by the camera system.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.