*3.4. Environmentally Triggerable Systems*

#### 3.4.1. Thermally Driven VFf

When the surrounding air is heated above the critical temperature of the SMA spring, temperature-induced phase transformation takes place, and the spring contracts and generates enough force to pull the ears downwards thereby closing the lobes. To ascertain the critical temperature of the SMA springs, they were stretched to 115 mm and placed in an oven. Between 60 ◦C and 65 ◦C, four of the five tested springs reduced their length by at least 50%. The spring that did not contract sufficiently was not tested further (see supplementary materials Figure S3). In some cases, these differences may originate from the production process, as these were commercially available SMA spring. Three springs with comparable contraction behavior were selected for further experiments with the VFf. These were attached to the VFf ears, and the closing behavior was investigated at three different temperatures (55 ◦C, 60 ◦C, and 65 ◦C) via video analysis and motion tracking (Figure 6A,D). No movement was detected at 55 ◦C, only one demonstrator closed its lobes at 60 ◦C, whereas all demonstrators performed closing movements at 65 ◦C (Figure 6B). The closing movement was triggered by a reduction in spring length, which was here measured at the beginning and the end of a heating cycle at the target temperature (Figure 6D lower orange lines). Spring contractions ranged from 0 cm to 5.44 cm (Figure 6B). A minimal spring length reduction of 0.56 cm was needed fully to close the lobes of the demonstrator (Figure 6B red line). The overall time needed from the first sign of movement to the complete closure of the VFf was defined as the closing time. The closing times of the three VFf ranged from 36 to 429 s with an average of 234.7 s (Figure 6C), the fastest closing times being observed at 65 ◦C. The average closing speed ranged from 0.19 <sup>×</sup> <sup>10</sup>−<sup>3</sup> to 1.1 <sup>×</sup> <sup>10</sup>−<sup>3</sup> m/s (Figure 6E). The force generated by the spring contraction was measured at 65 ◦C. Spring two and three showed no significant difference with an average force between 4.17 N and 6.83 N (Figure 6F). The corresponding raw data for spring length change is shown in Table S5 and for demonstrator closing times in Table S6 in the supplementary material section "Materials and Methods: 7. Environmentally triggerable systems: Thermally driven AVF using shape memory ally (SMA) springs".

3.4.2. Motion Analysis in VFfs Actuated by a Combination of Two Stimuli: Humidity and Temperature

The change in curvature of the hydrogel-coated backbones under changing environmental conditions was analyzed in a small climatic chamber that allowed controlled changes to be made in air humidity and temperature. Seven hydrogel-coated backbones were tested at a constant temperature of 65 ◦C, and the bending angle was observed at various humidity levels (Figure 7A). All seven backbones straightened completely at 60% relative humidity. The process started slowly, and minor changes became visible at 25% to 35% humidity. Between 35% and 60%, the decrease in the opening angle accelerated, levelling out at 65% to 80% humidity (Figure 7B).

After attachment of the hydrogel coated backbone to the foil VFf, a higher environmental humidity of about 70% was needed to reduce the curvature of the backbone and thus to unlock the demonstrator. At higher humidity, the hydrogel swelled to a greater extent, as was necessary to overcome the resistance of the foil material. After the VFf had been placed in a stream of hot steam (Figure 7C), it took an average of 64.5 s to unlock and reach the initial resting position for manual actuation (Figure 7D). Out of 95 tests, 62 demonstrators unlocked correctly during the 300 s rehydration time in hot steam. The average time required from the first contact of the steam to the unlocking of the VFf varied from 33.2 s to 126 s during the five cycles, whereas the actual unlocking movement (*n* = 19) required approximately 0.404 s (sd = 81.4 ms).

*Biomimetics* **2022**, *7*, x FOR PEER REVIEW 15 of 27

**Figure 6.** Kinematic motion analysis of the thermally driven VFf. Each demonstrator was actuated **Figure 6.** by a thermo-responsive SMA spring Kinematic motion analysis of the thermally driven VFf. Each demonstrator was actuated , which contracted and thus performed a closing motion when by a thermo-responsive SMA spring, which contracted and thus performed a closing motion when a critical temperature was reached. The tested demonstrators were placed in a climate chamber at three different temperatures. Videos were recorded from two perspectives (**A**,**D**), and the closing movement was further analyzed via motion tracking. Spring length was determined before and after the heating cycle (**D**), and the difference in length was calculated for complete closure. In (**D**), the upper orange line represents the demonstrator length and lower orange line the spring length before actuation. All boxplots above the red line indicate full closure (**B**). Each boxplot represents seven measurements. Different letters represent significant differences from other results with *p*-values below 0.05. Closing times (**C**) and speeds (**E**) were measured with the help of the motion paths (colored tracks of lobe tips). Closing times ranged from 36 s to 429 s with an average of 234.7 s, whereas average closing speeds ranged from 0.19 <sup>×</sup> <sup>10</sup>−<sup>3</sup> m/s up to 1.1 <sup>×</sup> <sup>10</sup>−<sup>3</sup> m/s. Each boxplot represents 14 measurements. In order to determine the force that was generated by the springs at closing temperature (65 ◦C), they were placed in the thermal chamber of the testing machine, and the output force was measured (**F**).

**Figure 7.** Autonomous environmentally triggered motion of VFf with a hydrogel-coated SMP backbone. In the dried state, the hydrogel-coated backbone is bent ((**A.1**), 20%) but straightens when the level of humidity increases ((**A.2**), 45%; (**A.3**), 80%). At a constant temperature of 65 °C and 60% humidity, the backbone reaches 0° curvature (**B**) *n* = 7. The straightening process of the locked demonstrator (with foil attached) needs higher levels of humidity to overcome the resistance within the bent foil. Approximately 70% humidity is required to unlock the hydrogel demonstrator. To determine the unlocking time precisely from contact to switching, a stream of hot steam was directed selectively towards only the hydrogel-coated backbone (**C**). The time needed for the unlocking of the demonstrator ranged from 16.8 s to 241 s after first contact with the stream of hot water vapor (**C**,**D**). Significance levels are indicated by the capital letters next to each boxplot. Different letters represent significant differences from other results with *p*-values below 0.05. In this testing scenario, 19 demonstrators were tested five times. If the demonstrator did not unlock within 300 s, the test was declared unsuccessful (**E**). **Figure 7.** Autonomous environmentally triggered motion of VFf with a hydrogel-coated SMP backbone. In the dried state, the hydrogel-coated backbone is bent ((**A1**), 20%) but straightens when the level of humidity increases ((**A2**), 45%; (**A3**), 80%). At a constant temperature of 65 ◦C and 60% humidity, the backbone reaches 0◦ curvature (**B**) *n* = 7. The straightening process of the locked demonstrator (with foil attached) needs higher levels of humidity to overcome the resistance within the bent foil. Approximately 70% humidity is required to unlock the hydrogel demonstrator. To determine the unlocking time precisely from contact to switching, a stream of hot steam was directed selectively towards only the hydrogel-coated backbone (**C**). The time needed for the unlocking of the demonstrator ranged from 16.8 s to 241 s after first contact with the stream of hot water vapor (**C**,**D**). Significance levels are indicated by the capital letters next to each boxplot. Different letters represent significant differences from other results with *p*-values below 0.05. In this testing scenario, 19 demonstrators were tested five times. If the demonstrator did not unlock within 300 s, the test was declared unsuccessful (**E**).

#### *3.5. Energy and Work/Kinetic Energy Requirements and Stored Energy for and during Lobe Movement*

The energy consumption of each experiment was determined by measuring the amount of electricity that was required for actuation. The electricity consumption was determined with an energy-measuring device that was plugged into the socket before the actual electricity consumer. Between 0.24 and 0.23 L of compressed air was needed for each actuation cycle of the pneumatic demonstrator, and the pneumatic test bench consumed 1 J per cycle (closing and opening). The magnetic stirrer consumed between 27 J and 39 J for the duration of actuation depending on rotation speed. Unlocking of the heat- and moisture-driven hydrogel actuator through a hot water steam source consumed 129,712 J, whereas the SMA-based actuation through contactless heating in a temperature chamber consumed 5,464,800 J of electric energy (Table 1). The efficiency of the systems is for pneumatic actuation around 3.7%, for magnetic actuation 5.6%, whereas for the SMA spring it is far lower as the temperature chamber used has a very high power consumption (Table 1).

Force-displacement measurements were performed to determine the amount of kinetic energy (work) required either to open or to close the VFfs (Figure 8). The closing movement of the pneumatic VFf needed an average amount of kinetic energy between 19.04 mJ and 30.81 mJ. Most of the time (97.3%), the opening motion of the two lobes was slightly timeshifted, as one lobe opened before the other (Figure 8C,D). For the opening movement of the VFf, an average kinetic energy requirement of 24.42 mJ to 34.65 mJ was needed to trigger the first opening event (Figure 8D,E). Between 38.49 mJ and 79.54 mJ were necessary for the second opening event (Figure 8D,E). A parallel or simultaneous lobe opening demanded less kinetic energy (around 37 mJ), although the force needed was higher (10 N compared with 9 N for the second lobe opening). The overall stroke necessary to achieve the opening movement was less than that for the non-parallel opening (Figure 8F). During opening, this difference was also reflected in the observed force drop, which was lower for non-parallel opening than for parallel simultaneous opening (Figure 8E). The release of the stored energy was reflected in the drop in force after the opening of a lobe. The released stored energy was calculated with Equation (1) to be approximately 0.6 mJ for the opening of one lobe and 2.4 mJ for the opening of both lobes in parallel.

Similar experiments were performed on the magnetic VFfs by applying a downward force on the magnets (Figure 8G,H). The mean kinetic energy requirements ranged between 11.63 mJ and 17.31 mJ. The measured curves for the closing movement of the pneumatically and magnetically driven VFf differed from each other. Both showed a linear increase in force at first, but for the pneumatic demonstrator, the curve became steeper at the end of stroke (B), whereas for the magnetic demonstrator, it flattened, and even decreased (H). The average force generated by the SMA springs at 65 ◦C ranged from 4.17 N to 6.83 N (Figure 6F) and the change in spring length (Figure 6B) from 5.6 mm to 54.4 mm. Based on these values, the average kinetic energy requirements computed to 48.1 mJ to 315.9 mJ. The energy values of all tested VFf systems in comparison with the biological model are shown in Table 1.

*Biomimetics* **2022**, *7*, 99


time: 33.2 s to 126 s

s

77%

steam production -

**Snap-Snap-Snap-Snap-**

**Kinetic Energy Re-**

**Kinetic Energy Re-**

**Energy Consump-**

**Energy Consump-**

**Efficiency** 

**Efficiency** 

**Re-**

**Re-**

**Table 1.** Comparison of *D. muscipula* with artificial Venus flytraps.

**Table 1.** Comparison of *D. muscipula* with artificial Venus flytraps.

**Table 1.** Comparison of *D. muscipula* with artificial Venus flytraps.

**Table 1.** Comparison of *D. muscipula* with artificial Venus flytraps.

**Table 1.** Comparison of *D. muscipula* with artificial Venus flytraps.

*Biomimetics* **2022**, *7*, x FOR PEER REVIEW 19 of 27

*Biomimetics* **2022**, *7*, x FOR PEER REVIEW 19 of 27

*Biomimetics* **2022**, *7*, x FOR PEER REVIEW 19 of 27

*Biomimetics* **2022**, *7*, x FOR PEER REVIEW 19 of 27

*Biomimetics* **2022**, *7*, x FOR PEER REVIEW 19 of 27

**Figure 8.** Kinematic energy measurements: examples of force displacement measurements carried out on the pneumatic and magnetic VFfs. Force measurements are used to calculate the energy needed to close and open the system. To imitate the closure of the pneumatic demonstrator, a piston with two pins is used to apply pressure to the backbone until the lobes close (**A1**,**A2**,**A3**). The force (**B**) necessary to close the lobes is measured repeatedly. The force for the opening movement is measured with a piston designed to apply pressure at the center of the backbone (**C1**,**C2**,**C3**). In most cases, the lobes open one after another (**C2**,**C3**) resulting in two distinct sudden drops in the force measured (**D**). When both lobes open simultaneously (**F**), the measured decrease in force (indicated by \* in (**E**,**F**)) is larger compared with the separate opening events (**E**). Each of the three demonstrators used for video analysis were tested 25 times for opening and closing movements, respectively. The magnetic demonstrator (**G**,**H**) needed less force compared with the pneumatic de-**Figure 8.** Kinematic energy measurements: examples of force displacement measurements carried out on the pneumatic and magnetic VFfs. Force measurements are used to calculate the energy needed to close and open the system. To imitate the closure of the pneumatic demonstrator, a piston with two pins is used to apply pressure to the backbone until the lobes close (**A1**–**A3**). The force (**B**) necessary to close the lobes is measured repeatedly. The force for the opening movement is measured with a piston designed to apply pressure at the center of the backbone (**C1**–**C3**). In most cases, the lobes open one after another (**C2**,**C3**) resulting in two distinct sudden drops in the force measured (**D**). When both lobes open simultaneously (**F**), the measured decrease in force (indicated by \* in (**E**,**F**)) is larger compared with the separate opening events (**E**). Each of the three demonstrators used for video analysis were tested 25 times for opening and closing movements, respectively. The magnetic demonstrator (**G**,**H**) needed less force compared with the pneumatic demonstrator and also less energy. The force-displacement curves shown in (**B**,**D**–**F**,**H**) represent exemplary curves. Significance levels are indicated by the capital letters next to each boxplot. Different letters represent significant differences from other results with *p*-values below 0.05 (**B**).

#### **4. Discussion**

We have created an environmentally triggerable biomimetic system combining two movement principles in one cost-efficient structure mimicking biological models. The

movement characterization with direct and precise pneumatic actuation has revealed that one motion cycle of our VFf system consists of five phases, beginning with a fast closure within 311 ms, followed by an even faster snap opening within 73 ms based on the release of stored elastic energy during opening. Through contactless, magnetically driven actuation, the VFf achieves an oscillating motion of the lobes near to the estimated resonance frequency of the system. The SMA spring and hydrogel-coated SMP backbone actuated VFfs highlight the possibility of an actuation of the system induced by environmental stimuli. Overall, the first combination of two snap-trap motion principles within one compliant system provides the baseline not only for novel artificial Venus flytraps, which are as fast as the biological model (pneumatic VFfs) and which act just as autonomously (3D-printed hydrogel-coated VFfs), but also for application in the fields of bioinspired architecture, autonomous systems, and soft robotic actuation.

High-speed recordings have allowed us to study the complete movement of the capture lobe of *D. muscipula*, an event that lasts on average less than 500 ms, matching well with the literature values (100–500 ms [6,27,29]). We have observed that, during a threeweek testing period, the plants tend to take longer and longer to close after an unsuccessful attempt at catching prey, a rest period, and a full reopening. We have been able to record and track the time and speed of the fast snap buckling motions in the biological model and to compare them with measurements of the motions of our VFf systems. With the presented characterization process and setup, it is possible to analyze fast 3D movements with high-resolution videos at 1000 fps frame rate.

#### *4.1. Comparison with the Biological Model*

Our VFf systems take less time for closure and generate higher speeds than the biological models (Table 1). Table 1 represents an updated extension of the table published by Esser et al. [13] and compares the individual VFf systems of this study with the biological model. To determine the energy required for closure, the energy consumption of each experiment was determined by measuring the amount of electricity that was required for actuation via an energy-measuring device. Force-displacement measurements were performed to determine the kinetic energy required for closure and the potential energy stored in the system during closure. The VFf systems required less energy (1 J in the case of pneumatic actuation derived from electrical energy consumption) for lobe closure than *D. muscipula*, which needed approximately 300 µmol ATP (ca. 10 J under standard conditions [27,28]). A comparison with other AVF literature values is difficult, since no energy measurements or calculations have been performed for most systems. The SMA spring-driven system of Kim et al. [30] needs 12.4 J for closing and 48 J for reopening, which are lower values than for our thermally driven SMA system and higher for the biological model. Their AVF system is faster in closing than our SMA-driven system. As we use a thermal chamber for heating instead of direct heating via electrical current (joule heating), our system heats up far more slowly than that of Kim et al. [30]. However, the former more clearly reflects heating attributable to a change in ambient temperature and validates the feasibility of our system for its intended use. In contrast to the system of Kim et al., the highest movement speeds in our study were achieved with the pneumatically and magnetically driven VFf systems.

#### *4.2. Combination of Two Snap-Trap Principles Gives a Novel Pneumatically Driven Motion Sequence*

The actuation time (AT) significantly influenced the closing time of the pneumatic VFf, with a shorter AT resulting in a shorter closure time. As the necessary actuation pressure inside the cushions was 0.7 bar, the overall system pressure had to be higher to achieve this value during shorter AT. The closed state of the pneumatic VFf represents a high energetic state, where energy is stored by elastic deformation within the backbone and by distortion of the lobe geometry. This energy can only be released suddenly by passing a threshold in pressure and increasing the curvature of the backbone, which thereby transforms the stored elastic energy into kinetic energy causing the observed rapid movement of the lobes. Thus, the pneumatic cushions are not involved in this opening process; as such, their AT has only marginal to no influence on the opening time. The small differences observed are attributable to variations in the acceleration of the lobe. Significant differences above 40 ms in the opening times are attributable to a delayed and non-parallel opening movement of the two lobes. The observed large differences are probably based on inaccuracies during the manufacturing process, as the demonstrators are produced by hand, which can lead to (small) geometric deviations between the individual demonstrators.

A comparable pneumatic AVF system from Pal et al. [14] consists entirely of silicone and is able to close its trap lobes within 50 ms when using 0.35 to 0.7 bar for actuation, comparable with the pressure used in our pneumatic VFfs. In contrast to Pal's system, our system stores energy during the closure movement within its lobe curvature similar to *D. muscipula* and does not rely on a pre-stretched layer. Therefore, the opening of our system shows actual snap buckling similar to that achieved in the biological model by releasing the stored energy in the lobes and thereby producing a rapid movement.

#### *4.3. Resonance-like Movement and Generation of Contactless Fast Flapping Motion*

Driven via a rotating magnetic field, the VFf is able to achieve a uniform oscillation or resonance-like motion depending on the rotational speed. Since the lobe motion is kinematically coupled to the ear motion, a significant influence of the different actuation speeds on the lobe motion in terms of velocity and overall motion has been observed. At 400 and 700 rpm, only minor motion of the VFfs is observed, but as the system approaches its presumed natural frequency at about 800–900 rpm based on acoustic and visual regularity, motion becomes more regular, and high flapping velocities and large amplitudes of motion are achieved. At more than 900 rpm, the movement becomes more irregular, and the lobes undergo kinking. Their movement could no longer follow the movement of the ear, which moves up and down faster than the lobes.

A comparison with the magnetic AVF from Zhang et al. [15,16] is difficult as their system uses an electromagnet to repulse a magnet attached to the AVF lobe, closing the system in 100 ms. Our system does not fully close its lobes and performs a flapping motion instead of closure.

#### *4.4. Environmentally Triggered Motion*

The closing of the SMA spring was significantly slower compared with that of the pneumatic VFf presented. Once the transition temperature was reached, the movement was continuous but slow. Notably, the used climate chamber showed effects of inevitable thermal stratification because of the low fan speed used during video recording (Figure S6). This led to the VFf nearest to the ceiling of the chamber reaching the critical transition temperature first. The systems showed no significant difference in closing speed or time (necessary for full closure) once the closing motion was initiated at transition temperature.

A direct comparison with the SMA-powered AVFs by Kim et al. (2014) is impossible, as the latter is based on bi-stable prepreg surfaces, in which the SMA springs are used to overcome the tipping point in the bi-stable system resulting in a fast-snapping motion, which takes from 0.4 s to 1.2 s [30].

Autonomy and environmental triggers are introduced into the system with the incorporation of shape memory materials. The VFf closure movement can be triggered by a change in temperature. This allows the system to be locked into a snapped open state until a specific stimulus combination unlocks the system and triggers the release of stored elastic energy. The blocking and locking of the movement and the specific release mechanism are achieved by the material properties of the components and the combination of shape memory polymers with hydrogels. We suspect that the observed differences between the individual test runs are attributable to the reduction and build-up of internal stresses in the system. The test conditions were not changed between the test runs, and the demonstrators were all dried in the same way. This creates an unwanted variance within the system, an issue that will be investigated in further studies. A comparable hydrogel-based system is

the hydroscopic bi-stable sheet AVF of Lunni et al. [21], which responds to a 30% increase in relative humidity. Our hydrogel requires an increase of 35% in combination with the SMP backbone to create movement.

#### *4.5. Overall System Discussion and Outlook*

The novel motion sequence achieved here by combining the movement principles of two snap-trapping plants uses motion amplification for closure and snap buckling for opening. The closure movement temporary stores energy via the deformation of the foil and backbone: through the bending of the attached backbone, the foil crease buckles, and the systems snaps open. This novel motion pattern can be used for various applications in energy harvesting, safety switches, escape movements of UAVs and bioinspired architecture. Triggering and actuating the motion can be achieved through various actuation principles.

Compared with the biological model and other AVF systems, only the pneumatically actuated VFf requires less energy for actuation and shows a similar kinematic behavior to the biological model with curvature reversal of the lobes and a fast snapping motion. Moreover, the system is easy and inexpensive to produce, by using office supplies (overhead foil, magnets, double-sided adhesive tape) and 3D-printed parts with a low heat deflection temperature in the range of 50 ◦C. Our systems provide an alternative to the current stateof-the-art AVF in terms of actuation and environmental sensitivity and allow various types of actuations to be tested in the same base module.

For example, the pneumatic actuation system represents a suitable actuator to achieve precise and controllable motions. In combination with the specific structure of the VFf, the novel movement and snapping mechanism builds a mechanical energy buffer. During the closure movement, potential energy is mechanically stored in the system and is explosively released during parallel opening. The attachment of triboelectric harvesters to the lobes might allow the partial recuperation of the kinetic energy of the snapping motion (release of potential energy during opening 2.4 mJ), similar to the leaf harvesters of Meder et al. [1,31].

By using different materials and material combinations, the principle shown can also be transferred to other application areas. The autonomous actuation depends on the properties of the materials employed and can be achieved by other environmentally triggered shape memory materials, e.g., liquid crystalline elastomers [20,32–34]. Important here is the geometry and the combination of flexible and rigid material, equivalent to those of the foil and backbone, which enable a snapping motion in the first place. The system can also be implemented using a multilayer film system, whose individual films have different coefficients of thermal expansion, strengths, and stiffness.

The combination of the present actuation strategies into one system is yet to be achieved. Two combination strategies are possible. First, the combination of pneumatic actuation and the hydrogel/SMP-based system initialization mechanism should enable quick opening and closing via pneumatics and the locking of the system in place by hydrogel drying without having to expend energy. The second possibility arises from combination of the thermally driven SMA spring with a hydrogel latch creating a latchmediated spring actuation (LaMSA) system [11,35]. The latch holds the spring in place, until an increase occurs in humidity. To function, the hydrogel-coated backbone must be stiff enough to withstand the contraction force generated by the SMA spring. This only applies if the system is heated above 65 ◦C, e.g., by direct sunlight, which causes the spring to contract and the SMP backbone to heat up beyond its TG. In future studies, flexible sensors will be integrated into the system, representing a first step in the development of a fully autonomous AVF system.

#### **5. Conclusions**

In this study, we present a novel motion sequence for plant-inspired robotics that combines, for the first time, the motion principles of two snap-trap mechanisms, as observed in plants, into an artificial Venus flytrap, namely the artificial Venus Flyflap (VFf) system. This system is based on a compliant foil with a rigid backbone resembling the snap-traps

of plants. A closing step via kinematic coupling and motion amplification of the lobes and a fast opening through snap buckling of the VFfs backbone characterize the motion. A uniform resonance-like motion is achieved via actuation through rotating magnetic fields. Two autonomous environmentally triggerable versions are presented based on shape memory materials in combination with a hydrogel coating that reacts to changes in temperature and humidity. In addition, a novel selective locking mechanism is introduced that only unlocks if two environmental stimuli (in this case, changes in humidity and temperature) are present at the same time. The VFf system is characterized in terms of motion kinematics, speed, times, and energy requirements and is compared with the biological model and state-of-the-art AVF systems, all of which highlight the biomimetic potential and low energy demand of our system.

In future studies, the presented environmentally triggered actuators will be merged into one system with self-healing foil materials systems. In addition, the integration of soft flexible sensors and flexible solar cells as energy harvesters will lead to an autonomous system with embodied intelligence and energy. Hereby, we should achieve an autonomous, adaptive, and self-healing artificial Venus flytrap system advancing soft plant-inspired robotics and bringing technology one step closer to the ingenuity found in naturally occurring systems.

**Supplementary Materials:** The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/biomimetics7030099/s1, Supplementary Material document including further information as well as the cited figures: Figure S1. The pneumatic test bench provided pressurized air for the demonstrator [25]. Figure S2. Setup for magnetic demonstrator testing. Figure S3. Frequency measurements of magnetic demonstrators. Figure S4. Determination of critical temperature and contraction behavior of the five SMA-springs. Figure S5. Climate chamber setup for SMA-spring demonstrator measurements. Figure S6. Temperature distribution inside the climate chamber at the different levels of the demonstrators. Figure S7. 3D printed SMP backbone for coating with hydrogel. Figure S8. Preparation of full assembly of environmentally triggerable systems stimulus combination system [36,37]. Figure S9. Determination of the opening angles and the torsion of the backbones. Table S1. Kinematic analysis of the Dionaea muscipula was performed with the same setup as the magnetic and pneumatic demonstrator. Table S2. Maximum force generated by SMA springs at 65 ◦C. Table S3. The actuation pattern, actuation time and actuation pressure that were used to achieve repetitive actuation are displayed. Table S4. Flapping frequency corresponding to different actuation speeds of the magnetic demonstrators. Table S5. Length change of SMA springs at different temperatures when mounted to a foil demonstrator. Table S6. Closing times of the SMA spring demonstrator at different temperatures [37].

**Author Contributions:** Conceptualization, F.J.T., F.D.S. and T.S.; methodology, F.J.T. and F.D.S.; formal analysis, F.J.T., F.D.S. and T.S.; investigation, F.J.T., P.A. and J.T.; data curation, P.A. and F.J.T.; writing—original draft preparation, F.J.T. and P.A.; writing—review and editing, F.J.T., F.D.S. and T.S.; visualization, P.A. and F.J.T.; supervision, F.J.T. and T.S.; project administration, F.J.T. and T.S.; funding acquisition, T.S. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under Germany's Excellence Strategy-EXC-2193/1-390951807.

**Data Availability Statement:** Data is available in the supplementary material document and on request.

**Acknowledgments:** F.J.T. and T.S. are grateful to the Deutsche Forschungsgemeinschaft for funding their research. The authors thank R. Jones for improving the manuscript language. The authors thank various colleagues within the framework of the cluster of excellence "Living Materials Systems (livMatS)" for inspiring discussions.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study, in the collection, analyses, or interpretation of data, in the writing of the manuscript, or in the decision to publish the results.
