Plants turn out in a perfect hydraulic engine. At the macro level, plants exploit a sophisticated strategy, relying on water potential gradient, for moving water from the soil (i.e., at soil level ≈ −0.3 MPa) to the leaves (i.e., at the leaf level ≈ −7.0 MPa; McElrone et al., 2013). At the cell level, plants draw the water in and out of their cells using the osmotic gradient across semipermeable membranes. The turgor pressure thus changes creating a local change in the cellular volume and tissue stiffness, and by exploiting the thin and stiff cell-wall features, enables the large-scale tissue deformations required for motion (Dumais and Forterre, 2012). The characteristics of the water flow induced by gradients of water potential, the level of turgor pressure, and the mechanical properties of cell-wall deformation, are the key elements of water-driven movements in plants.

In roots, the movement of primary growth is characterized by the expansion of the cells facilitated by water uptake generating turgor pressure to inflate the cell and stretch the walls (Taiz and Zeiger, 2002). Plants are thus able to penetrate and explore soil in a non-destructive way. Such mechanism was first investigated from a robotics point of view in Mazzolai et al. (2011), with an osmotic actuation module implementing electro-osmosis by three cells separated by pairs of semipermeable osmotic membranes and ion-selective membranes, individually coupled with a piston mechanism. To aid the robotic design, Sinibaldi et al. (2013) followed a bioengineering approach to model the dynamics of osmotic actuation and represent a formal expression of scaling laws for the physical parameters necessary for the actuation strategy: characteristic time, maximum force, peak power, power density, cumulative work and energy density, role of volume-surface aspect ratio. This model was then exploited to design a forward osmosis-based actuator (Sinibaldi et al., 2014), fabricated on the basis of an analysis of plant movements, plant cell characteristics, and osmotic actuation modeling. The system has a typical size of 10 mm, produces forces above 20 N, with a power consumption in the order of 1 mW, and a characteristic time of 2–5 min.

To obtain faster movements, plants exploit instability and fluid-solid coupling together with hydraulic mechanisms: elastic energy is first stored in the cell walls by means of a water flow, and released suddenly when a critical threshold of energy barrier is surpassed (Forterre et al., 2005; Dumais and Forterre, 2012). The rapid movement of leaf closure in the carnivorous plant Dionaea muscipula (Venus flytrap) derives from the accumulation of elastic energy in the leaves, driven by swelling and shrinkage coupled with a double curvature geometry of lobes. This allows a snap-buckling mechanism of 100 ms after the initial trigger stimulus. These features inspired a flytrap-like robot described in Kim et al. (2014) which can reach rapid speed motion (~100 ms) and large deformations (18 m⁻¹). The system consists of an asymmetrically laminated carbon fiber prepreg (CFRP), which acts as a bistable artificial leaf, and a shape memory alloy (SMA), which acts as triggering actuator to induce the snap motion. Differently, Esser et al. (2019) combined different actuation systems (pneumatic, plus SMA spring) to translate the principles of movements of Venus flytrap and waterwheel plant in systems able to response to different environmental triggers (heat, moisture or magnetic stimuli). More recently, polarized light microscopy revealed the presence of microfibrils reinforcing the leaf, running perpendicular to its midrib in the upper and lower epidermis. This additional feature, integrated with bistability, inspired the design of an artificial hygroscopic bistable system, obtained by bonding prestretched poly(dimethylsiloxane) – PDMS layers prior to depositing electrospun polyethylene oxide (PEO) nanofibers (Lunni et al., 2020). The Venus flytrap is an interesting model for robotic components and artificial materials, which have been recently deeply reviewed by Esser et al. (2020).

The movement of roots inside the soil occurs by adding new cells to the apex. This strategy allows minimizing the resistance forces during penetration, and is helped by lateral hairs and diameter expansions that keep the whole system anchored. The roots' strategy of growing from the tip is a key specification for the development of robotic systems for soil exploration. To quantify the influence of this mechanism during penetration, Tonazzini et al. (2013) used a physical robotic demonstrator and showed that the growth from the tip reduces penetration energy from 20 to up 50%, depending on the initial depth. Following these quantitative analyses, Sadeghi et al. (2014) designed the first root-like system implementing a growing mechanism by means of a monotonic process that continuously adds new material to the base of the tip, in the form of a layer, and pushes forward the tip itself layer-by-layer.

This concept triggered the idea of moving-by-growing, paving the way for growing robots: by integrating a miniaturized 3D printer into the tip of a root-like device and using a thermoplastic filament, the body of the root-like system can be created layer-by-layer thus replicating the natural mechanism of cell deposition and consolidation that occurs in plants growth (Sadeghi et al., 2017). The concept of growing system has been approached also using other technologies. For example, the robot described in Hawkes et al. (2017) grows via pressurization of an inverted thin-walled vessel.

To further enhance exploration abilities, circular movements (known as “circumnutations”) are performed by the root tip due to a combination of internal factors and external factors (i.e., gravitropism, Brown, 1993; Stolarz, 2009; Migliaccio et al., 2013). Circumnutations are a class of movements that are found in all plants organs, but in particular in those that are involved in growth (roots, shoots, branches, flower stalks) and generate elliptical or circular trajectories (Mugnai et al., 2015).

To quantify the characteristics of root circumnutations in soil for robotic design purposes, Popova et al. (2012); Del Dottore et al. (2018b), and Tedone et al. (2020) have proposed a methodology for the analysis of the movement, including a time-lapse videos observation (in air and in soil) and the study of tip kinematics using the “Analyser for Root Tip Tracks” (ARTT, Russino et al., 2013), which combines a segmentation algorithm with additional software imaging filters in order to realize a 2D tip detection. Measurements of the growing speed and circumnutation amplitude were extracted to implement the circumnutation behavior in a soft robotic root that bends using soft spring-based actuators (Sadeghi et al., 2016). Experiments in the air with the robotic root were performed to demonstrate that the system can follow an external stimulus while performing a circumnutation movement, similarly to a natural root. Additional experiments have been performed in a soil-like testbed showing that the use of plant root-like circumnutations improves the efficiency of penetration (33%) compared to moving directly forwards with no circular movement. In line with these results, additional investigations using “robophysical” modeling have revealed the benefits of tip nutation movements for navigating obstacles and exploring heterogenous terrains (McCaskey et al., 2019; Taylor et al., 2020).

To better understand the role of morphology of roots in soil, high-resolution imaging methods, such as micro-CT (Kaestner et al., 2006; Tracy et al., 2010) can be used to investigate natural roots and thus obtain a 3D reconstruction for bioinspired design insights. Mishra et al. (2018) used imaging capture of Zea mays roots via an optical microscope. The aim was to extract the morphological features of the tip profile and implement a 3D CAD model to guide the design and fabrication of 3D printed root-like probes. These devices, with different diameters and shapes, were compared in terms of energy consumption and penetration force via experimental tests in real soil and discrete element simulations, demonstrating the higher penetration performance of the bioinspired tip profile with respect to the other ones.

Early stem growth, where young shoots extend into spaces and search for support, are known as “searchers.” An outstanding example of a light-mass searcher can be found in the climbing catus Selenicereus setaceus (Soffiatti and Rowe, 2020). Searchers often have a light but stiff structure, and are capable to extend across voids and perform circumnutations to improve the probability of touching a support (Gallenmüller et al., 2004). Circumnutations offer a valuable solution for adaptation algorithms in motion planning (Wooten and Walker, 2016, 2018). By analyzing the behavior of vines, circumnutation-based algorithm improves the performance of a tendril-like continuum robots, enabling efficient environmental contact and helping to guide and stabilize the system. Such strategy could be used, for example, in space for positioning sensors or exploring the surrounding environment (Mehling et al., 2006; Tonapi et al., 2014; Wooten and Walker, 2015; Nahar et al., 2017).

In climbing plants, contact coiling starts when a local mechanical stimulus occurs in a tendril, which then start to curl around the support and grip to it. This coiling is associated with the presence of gelatinous fibers (“G fibers,” Bowling and Vaughn, 2009). A differential decrease in the water content of a G fibers-bilayer ribbon generates an asymmetric contractile force, which drives the coiling (Gerbode et al., 2012). Then, a secondary coiling (“free-coiling”) pulls the plant closer to the support, creating an elastic spring-like anchorage resistant to external loads or wind. The loss of water during the free-coiling phase leads to an increase in structural rigidity, or lignification, which prevents it from uncoiling. Contact and secondary coiling were investigated from a kinematic viewpoint in Vidoni et al. (2015) to design and develop a tendril-like system able to grasp-by-coiling.

In some cases, the natural mechanism of coiling is reversible, so if the support is not suitable, the tendril uncoils. This feature was implemented in a small-scale system by Must et al. (2019). The actuation strategy derives from a plant's capacity to actively control osmolyte gradients, and is based on the electrosorption of ions on flexible electrodes (in porous carbon), driven at low input voltages (1.3 V). A 1 cm electroactive unit reversibly controls the concentration of ions (acting as osmolyte) obtained through the dissolution of an electrolyte. This tendril-like soft robot, coupled to the control unit, performs coiling and uncoiling with reversible stiffening and actuation.