*Article* **Novel Motion Sequences in Plant-Inspired Robotics: Combining Inspirations from Snap-Trapping in Two Plant Species into an Artificial Venus Flytrap Demonstrator**

**Falk J. Tauber 1,2,\* ,† , Philipp Auth 1,†, Joscha Teichmann <sup>1</sup> , Frank D. Scherag 2,3 and Thomas Speck 1,2,4**


**Abstract:** The field of plant-inspired robotics is based on principles underlying the movements and attachment and adaptability strategies of plants, which together with their materials systems serve as concept generators. The transference of the functions and underlying structural principles of plants thus enables the development of novel life-like technical materials systems. For example, principles involved in the hinge-less movements of carnivorous snap-trap plants and climbing plants can be used in technical applications. A combination of the snap-trap motion of two plant species (*Aldrovanda vesiculosa* and *Dionaea muscipula*) has led to the creation of a novel motion sequence for plant-inspired robotics in an artificial Venus flytrap system, the Venus Flyflap. The novel motion pattern of Venus Flyflap lobes has been characterized by using four state-of-the-art actuation systems. A kinematic analysis of the individual phases of the new motion cycle has been performed by utilizing precise pneumatic actuation. Contactless magnetic actuation augments lobe motion into energyefficient resonance-like oscillatory motion. The use of environmentally driven actuator materials has allowed autonomous motion generation via changes in environmental conditions. Measurement of the energy required for the differently actuated movements has shown that the Venus Flyflap is not only faster than the biological models in its closing movement, but also requires less energy in certain cases for the execution of this movement.

**Keywords:** plant-inspired robotics; artificial Venus flytrap; motion sequence; biomimetics; bioinspiration

### **1. Introduction**

Within the last decade, plant-inspired robotics has become established as a new emerging field of soft robotic science. One outstanding example is represented by the plant-inspired growing robots developed by the group at the Italian Institute of Technology; these robots have roots, tendrils, and leaves like their biological models and are able to grow, forage, and harvest energy from the environment [1–5]. A focus area within this field is the development of artificial systems with certain characteristics that are able (1) to exist autonomously, (2) to sense, adapt, and react to the environment, (3) to sustain their homeostasis by harvesting energy, (4) to sense damage, and (5) to possess self-repair functions. Such systems are currently still in their infancy but should revolutionize the technical world in the future.

The Venus flytrap (*Dionaea muscipula*) provides a suitable biological model for plantinspired robotics. These plants perceive their environment and adapt and react to it, all

**Citation:** Tauber, F.J.; Auth, P.; Teichmann, J.; Scherag, F.D.; Speck, T. Novel Motion Sequences in Plant-Inspired Robotics: Combining Inspirations from Snap-Trapping in Two Plant Species into an Artificial Venus Flytrap Demonstrator. *Biomimetics* **2022**, *7*, 99. https:// doi.org/10.3390/biomimetics7030099

Academic Editor: Jinyou Shao

Received: 15 June 2022 Accepted: 15 July 2022 Published: 22 July 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 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 (https:// creativecommons.org/licenses/by/ 4.0/).

of which are features required by robots. In addition, Venus flytraps carry out one of the most complex decentralized controlled process sequences in nature, namely a fast snap-buckling prey capture movement, which includes an upstream mechanical memory and several subsequent processes depending on successful prey capture [6]. The Venus flytrap can be considered as a bi-stable system, with the lobes being stable in a convex and concave configuration [6–8], changing through a snap-buckling motion similar to technical bistable systems [9–12]. The transference and combination of the basic principles of the Venus flytrap movement sequence has enabled the production of life-like artificial Venus flytraps (AVF) [13]. However, artificial Venus flytraps have so far been produced more as a by-product of the development of new actuation systems. AVFs have been used as suitable demonstrators of the capabilities of actuators in the development of novel pneumatic systems utilizing instabilities in elastic energy storage [14], magnetically driven bi-stable prepregs [15,16], ionic electroactive polymer metal composites (IPMCs) [17,18], light responsive liquid crystalline elastomers (LCEs) [19,20], and hygroscopic bistable systems (HBS) reacting to changes in humidity [21]. However, none of these AVF systems incorporates all the functions of the biological model *D. muscipula*. Most of them only exhibit a reasonably fast (sometimes without snapping) closure mechanism following an external stimulus, as Esser et al. [13] have shown in their review of the current state of AVFs. Nevertheless, a simple and low cost AVF with a novel motion pattern inspired by two carnivorous plants has been developed, inspired by those AVF systems: the Venus Flyflap (VFf). It uses various actuation modes and allows the investigation and characterization of the energy needed for actuation [22]. In addition, for the first time, this system combines motion features of two closely related snap-trapping carnivorous plants into one system, namely the snap-buckling of *D. muscipula* (Figure 1A,B) and the kinematic coupling and motion amplification of the Waterwheel plant (*Aldrovanda vesiculosa*) (Figure 1C,D) [6,7,23,24]. The trap closure movements based on these motion principles are among the fastest movements in the plant kingdom: *A. vesiculosa* needs 0.02 to 0.1 s [23] and *D. muscipula* 0.1 to 0.5 s [6] to close their trap lobes.

The compliant foil structure used in the VFf is based on a shape inspired by the biological models. A rectangle forms the base (analogous to the leaf midribs of *A. vesiculosa* and *D. muscipula*); attached to its long sides, two triangles represent the trap lobes, and two circles ('ears') at the short ends of the rectangle allow the actuation of the system through kinematic coupling (Figure 1E–G) [14]. A downwardly directed motion of the 'ears' results in motion amplification causing the fast closure of the system by a continuous motion, as found in *A. vesiculosa*. To be able to snap, the foil 'midrib' is reinforced with a plastic microscopic slide as a backbone. When force is applied centrally, the backbone changes its curvature, and the VFf opens its lobes in a fast snapping motion similar, but inverse in direction, to the snapping by curvature inversion in *D. muscipula*. Thus, a new motion sequence has been achieved by combining the motion principles of the two snap-trap plants into one artificial system. A fast (continuous) closing step, followed by a sudden release of the stored potential elastic energy in the system, causes a snap opening, as soon as the input energy exceeds a certain threshold.

The present study aims to answer the following questions. Is it possible to build a biomimetic system that combines and utilizes two movement principles in one cost-efficient structure, that mimics the biological models, and that can be triggered by environmental stimuli?

net varying the actuation speed is used to achieve and investigate a uniform energy-effective oscillation or resonance-like motion (Figure 1I). Concerning autonomous systems, the VFf can be also actuated by environmental stimuli such as heat and humidity or a combination of both, by using environmentally sensitive materials such as shape memory alloys and polymers in combination with hydrogels. Furthermore, a comparison of our system with the biological models and the state of the art in artificial Venus flytraps allows an

evaluation of its biomimetic potential.

**Figure 1.** After being triggered, the Venus flytrap (*Dionaea muscipula*) closes its lobes by inverting the curvature of the lobes from, as viewed from the outside, concave (**A**) to convex (**B**) through the **Figure 1.** After being triggered, the Venus flytrap (*Dionaea muscipula*) closes its lobes by inverting the curvature of the lobes from, as viewed from the outside, concave (**A**) to convex (**B**) through the release of stored elastic energy. In the waterwheel plant (*Aldrovanda vesiculosa*) (**C**), water displacement in turgescent cells lying along the midrib combined with the release of stored elastic energy results in bending of the midrib, which is kinematical amplified and causes trap closure (**D**). The compliant foil demonstrator, the Venus Flyflap (VFf) (**E**), combines both mechanisms to close and open its artificial lobes. When force is applied to the ears (**F**,**G**), the lobes close via a kinematic coupling mechanism. Force applied to the backbone triggers a fast-snapping opening movement similar to the closing movement of the Venus flytrap. These mechanisms have been incorporated into various actuation scenarios. Pneumatic cushions drive the motion and enable the characterization of each phase of the motion sequence (**H**). A resonance-like, rapidly oscillating, flapping motion is achieved by a magnet that is attached to the ears and that responds to a rotating magnetic field (**I**). The VFf can also be actuated by environmental stimuli, when fitted with shape memory alloy (SMA) springs (**J**) or a combination of environmentally sensitive materials such as SMA and polymers in combination with hydrogels reacting only to a change in humidity and temperature (**K**).

Direct actuation by means of pneumatic cushions enables the opening and closing movements of the VFf to be controlled in a targeted manner (Figure 1H). Thereby, specific aspects of the motion and system as a whole can be investigated separately. This should provide new insight into the motion behavior (closing and opening times and speeds), the kinematics, and the elastic energy storage of the system. Contactless actuation via a magnet varying the actuation speed is used to achieve and investigate a uniform energy-effective oscillation or resonance-like motion (Figure 1I). Concerning autonomous systems, the VFf can be also actuated by environmental stimuli such as heat and humidity or a combination of both, by using environmentally sensitive materials such as shape memory alloys and polymers in combination with hydrogels. Furthermore, a comparison of our system with the biological models and the state of the art in artificial Venus flytraps allows an evaluation of its biomimetic potential.

#### **2. Materials and Methods**

### *2.1. Setups for Movement Analysis of Dionaea Muscipula and the Artificial Venus Flyflap (VFf)*

In order to enable a valid comparison of the motion characteristics of the *D. muscipula* plant with the pneumatically and magnetically actuated VFf, a kinematic analysis was performed using a video chamber fitted with two 1000 fps high-speed cameras (Baumer matrix monochrome camera VCXU 13 M/Imaging Solutions Motion traveller 1000) (Figure 2A). Videos were recorded at a constant frame-rate of 1000 fps with a resolution of 512 × 512 pix, and the recordings were synchronized using NorPix-StreamPix 8.0.0.0 (x64) Software. During actuation, the path of motion, movement speed, and kinematic parameters (speed and acceleration) were tracked and analyzed using the open-source software Kinovea (version 0.9.1). During the testing period of three weeks, 28 traps from four different *D. muscipula* plants were tested repeatedly. The tests were performed on three different dates with one week of "rest" in between to minimize the stress for the plants. Tracking markers were applied to the lobe tips of the pneumatic VFf and the magnetic VFf and to the magnets. The video and picture-based analysis of the thermally driven VFf actuated with shape memory alloy (SMA) springs (Figure 2B) and of the hydrogel actuated VFf (Figure 2C) required only a lower framerate because of the lower motion speed. Therefore, a digital camera (Panasonic Lumix DMC-FZ1000, Figure 2(C3)) with a recording framerate of 25 fps was used. In the case of the thermally driven VFf, the lobe tips and the spring length were tracked. Statistical data analysis was performed with the open-source software RStudio (version 1.2.5042).

The setup of the actuator systems is described in detail in Esser et al. [14] and shown in Figure 1. Further information is provided in the supplementary material section "Materials and Methods: 2.1 Standardized production of the compliant foil demonstrators". *Biomimetics* **2022**, *7*, x FOR PEER REVIEW 5 of 27

**Figure 2.** *Cont*.

**Figure 2.** Characterization setups. High-speed videos were recorded in a specifically build recording chamber (**A**) with two highspeed cameras (**A1**) positioned at a 90◦ angle to each other. Light is provided by two high-power flicker-free LED light sources (**A2**) that point at the object of interest (**A3**). The Venus flytrap and the pneumatic and the magnetic demonstrators were recorded using this setup.The SMA actuated demonstrators were tested by placing them in a temperature chamber (**B1**). Three demonstrators were tested simultaneously (**B2**). The movement of the springs and the lobes of the demonstrator were recorded by two cameras (**B3**) for later analysis. For actuation of the hydrogelbased demonstrator (**C1**), a hot water vapour source (**C2**) provided enough humidity to unlock the demonstrator, which was filmed by a camera for kinematic analysis (**C3**). Force-displacement measurements were performed with specific test mounts by the Hegewald and Peschke Inspekt Table 5 (**D**). The output force for each SMA spring was measured (**E1**) at 63 ◦C. To mimic the actuation procedure, specific mounts were designed to apply pressure in the direction of actuation in the demonstrator (magnetic (**E2**) and pneumatic (**E3**)). An exemplary force displacement curve of an opening event of the pneumatic system is shown in (**F**). S<sup>1</sup> and S<sup>2</sup> indicate the lower and upper end, respectively, of the integral (green area) used for calculating the kinetic energy necessary to open the the two lobes (lobe openig indicated in the curve by the two sudden force reductions).

#### *2.2. Energy Measurements*

The energy consumption in 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, e.g., the pneumatic test bench, magnetic stirrer, temperature chamber, or steam source. Force-displacement measurements were performed for calculating the necessary work and kinetic energy to drive and trigger the closing and opening movements of the demonstrators (Figure 2F). A specific testing setup to apply force according to the actuation scenario was designed for each VFf system (Figure 2E). The pneumatic VFfs were fixed in an upside-down orientation in the testing machine, and the compression pistons applied force to the backbone through the casing openings for the pneumatic cushions (Figure 2(E3)). In the case of the magnetic VFf, a downward force was applied to the magnets until the lobes closed (Figure 2(E2)) Clamps for the tensile tests held the SMA spring of the thermal VFf inside the thermal chamber of the testing machine. The temperature was increased up to 65 ◦C, and the output force was measured. All measurements were performed with an Inspect universal testing machine (Hegewald and Peschke Meß-und Prüftechnik GmbH, Nossen, Germany). During all hysteresis measurements, the stroke was adjusted accordingly to achieve a full closing or opening movement. The movement of the machine caused further increasing forces after a complete opening, since the testing machine did not immediately change the direction of movement after opening. These measurement ranges were excluded from the kinetic energy calculations. Since the course of the force was not uniform, it had to be integrated over the entire path until the moment when both trap lobes opened (*s*2) (Figure 2F). The area under the force-displacement curve represents the work exerted on the moving object. By measuring the area, the work and thus the required kinetic energy can be calculated by:

$$\mathcal{W} = \int\_{s\_1}^{s\_2} \stackrel{\rightarrow}{F} \left( \stackrel{\rightarrow}{s} \right) \* d \stackrel{\rightarrow}{s} . \tag{1}$$

with *W* = work/kinetic energy; *F* = force; *s* = stroke; *s*<sup>1</sup> and *s*<sup>2</sup> indicate the lower and upper end of the integral, respectively. The efficiency calculation and corresponding equations are shown in the supplementary material section "Materials and Methods: 9. Actuation system efficiency calculation".

#### *2.3. Motion Analysis of Discrete Repetitive Motion Generation by Pneumatic Actuation*

Analysis of the complex motion pattern of the VFf was achieved by using directly driven pneumatic actuation of the opening and closing motions, which could be triggered separately. The foil-based VFf was attached to a 3D-printed (RigurTM; Stratasys Ltd., Eden Prairie, MN, USA) ridged case that housed three pneumatic cushions (EcoFlex 0030; KauPo Plankenhorn e.K., Spaichingen, Germany) (Figure S1). Pressurization of the two smaller outside cushions pushed the rigid backbone of the VFf upwards, because its ears were attached to the casing (Figure 1H). This closed the lobes via kinematic amplification, as seen in *A. vesiculosa*. To trigger the opening movement, the central cushion inflated, while the outside cushions deflated. This caused the backbone to bend, leading to a fast "snapping" opening movement accompanied by an inversion of the spatial curvature of the lobes, similar to the snap-buckling principle of *D. muscipula*. Pressurized air was supplied by a pneumatic test bench [25] that enabled measurement and manipulation of the pressure and pressurization time for the closing and opening motions, respectively (see Table S3 for actuation pattern). As pressurization inflated the cushions and actuated the system, we further refer to the duration of pressurization as the actuation time (AT). The actuation times chosen for the characterization were 200 ms, 300 ms, and 400 ms, as the closing time of the biological model was below 500 ms. During actuation, the valves opened depending on the AT pressurizing the cushions. The actuation pressure was determined to be 0.7 bar in preliminary tests. This was the maximum pressure at which complete closure was

guaranteed without the cushions bursting. The system pressure was adjusted accordingly to the AT in order to achieve a pressure of 0.7 bar within the actuators during each AT.

#### *2.4. Motion Analysis of Contactless Actuation of the Demonstrator by a Rotating Magnetic Field*

In the magnetic VFf, the foil demonstrator was directly attached to a 3D-printed casing without the rigid microscopy slide backbone. The casing in turn was attached to an aluminum profile in the video chamber (see Figures 1I and S2). A set of two round flat magnets (Neodymium, 15 × 2 mm, 30 g each, 6 kg holding force) was attached to one of the ears. The VFf was placed 20 mm above a magnetic stirrer (IKA RCT B 5000) with adjustable rotations per minute (rpm) ranging from 100 to 1500 rotations per minute. The rotating magnet pulled and pushed the VFfs magnets, resulting in upwards and downwards movements of the VFf ear, which in turn set the lobes into a flapping motion. To investigate the possibility of reaching the natural frequency and gaining a resonance effect in the oscillating flapping motion of the system, six different actuation speeds were investigated with 400, 700, 800, 900, 1000, and 1300 rpm, respectively. In preliminary tests, the rpm were steadily increased from 0 to 1500 and the behavior of the VFf was observed. From 400 rpm onwards, the first clear movements occurred, at 800 and 900 rpm a uniform movement behavior was observed and for frequencies above 1300 rpm only irregular behavior. Because of this, the focus of the experiments was on the range of 800 and 900 rpm, and additional frequencies 100 rpm higher and lower were investigated. In addition, the extremes of 400 rpm (first clear movements) and 1300 rpm (only irregular erratic motions) were investigated.

#### *2.5. Environmentally Triggerable Systems*

#### 2.5.1. Motion Analysis of Thermally Actuated VFf by Using SMA Springs

To introduce autonomy into the system, the VFf was equipped with a temperaturesensitive SMA spring (Nitinol). The critical temperature for the SMA springs to induce contraction was determined by placing five springs in the temperature chamber of the universal testing machine at increasing temperatures, starting at 22 ◦C, and the length was measured each time that the temperature rose by 2 ◦C (see Table S2 and Figure S4). An SMA spring was attached to the foil demonstrator by using rivets on each lobe (see Figure 1J). The spring was stretched to a fixed length of 115 mm and then attached to the VFf. Three VFf were tested simultaneously in a climate chamber (Environmental test chamber CTC256, Memmert GmbH + Co. KG) that allowed programmed temperature settings (Figure S5). Three different temperatures were tested (55 ◦C, 60 ◦C, 65 ◦C) (Figure S6). The rising ambient temperature induced a phase transformation in the spring material reversing the deformation of the spring, which then shortened and, hence pulled the ears downwards closing the VFf.
