Introduction

Cyborg insects combine biological organisms with electronic elements and have the potential to surpass traditional robots in maneuvering through complex terrains1, enabling access to hard-to-reach areas. Insects embody a remarkable fusion of biological adaptations, allowing them to thrive across diverse habitats. Their ability to adapt to different environments highlights their low-maintenance nature2. With their distinct limb arrangement and positioning, they possess unparalleled means of interacting with their surroundings3. Certain species exhibit exceptional precision in flight maneuvers4, while others demonstrate swift recovery from injuries or trauma5, even regenerating lost limbs6. These characteristics are milestones we seek to achieve in the field of robotics.

Gromphadorhina Portentosa, commonly known as the Madagascar hissing cockroach, is a frequently employed species in cyborg insect research. This choice is attributed to its documented success in integrating electronics for steering its movements7,8,9,10,11,12. Due to its proven efficiency, it has been theorized that transforming this insect into a cyborg could be used as a reconnaissance tool in the event of a natural disaster.

In Japan, a nation highly susceptible to seismic events, search and rescue teams have invested in the development of advanced tools to expedite the rescue of earthquake survivors13. However, despite the integration of cutting-edge technology, rigorous training, and collaboration with service dogs, the availability of specialized personnel remains limited, thereby constraining their capacity to provide comprehensive assistance to a large number of victims. To address these challenges, the implementation of cyborg insects as a novel solution for conducting specialized search and rescue operations over extensive areas, with a reduced reliance on direct human supervision, has been proposed by several research groups14. The swift response of rescue teams plays a critical role in significantly improving the chances of survival for victims15,16,17,18,19.

When considering the altered locomotion characteristics of cyborg insects in the context of a disaster situation, it becomes evident that their innate terrain adaptability20 is constrained by the imbalance introduced by the electronic board, commonly referred to as the “backpack”. Incorporating electronics into a living organism significantly impacts its natural motion21,22,23. While some sensing elements and electronics can be easily adapted to smaller insects24,25, others, such as thermal cameras or radars, have elemental size and weight limitations12.

The Madagascar hissing cockroach is able to carry a load of up to three times its body weight. However, increasing the payload will proportionally affect its motion. A heavier load displaces the center of mass upwards, making it difficult for the insect to maneuver through terrain. When the payload exceeds the weight of the insect, the center of mass can be found outside the body, causing the insect to fall and land on its back when maneuvering in complex terrains.

We consider that to make an impact in disaster area reconnaissance. Cyborgs should be able to overcome obstacles and move freely. For this endeavor, we observed various insects self-righting locomotion and found out that winged beetles have an advantage in self-standing due to their additional limb which they can use to push themselves26,27.

We observed that the ladybird insect Coccinellidae used its wings as a last resort to escape from upside-down situations, with higher success over other insects with similar morphology. This phenomenon has been previously cataloged by Zhang et al. in their study focused on the self-righting of ladybirds27. The elytra, a robust chitin structure, safeguard their more delicate inner wings28, and it is the organ that actually pushes the insect. The combination of the elytra geometry and its coupling generates an oblique movement, which results in a rolling motion. This morphology has been previously implemented for the self-righting of fixed-wing drones29.

By leveraging the morphological characteristics of ladybird limbs into a bio-inspired artificial limb, we aim to enhance the capability of cyborg insects to overcome situations where they become immobilized, thereby enhancing the prospects of successful rescue missions.

Given the unpredictable nature of obstacles encountered in disaster situations, our research has placed emphasis on two key aspects: system adaptability and compact size. We have designed an active assistance mechanism for self-righting cyborg insects. The system offers a repeatable smart function that dynamically adjusts the wing position based on the prevailing situation, thus mitigating the risk of the cyborg insect getting stuck. The artificial limb creates an interface between the backpack and the insect surroundings.

In this study, we present comprehensive results and intricate details regarding the fabrication and assembly of the 3D printed mechanism and its electronic control. Our goal is to contribute to the advancement of future endeavors focused on developing advanced miniaturized components tailored for cyborg insects and the assistance of rescue services in disaster situations.

Results and discussion

Bio-inspired artificial wing design

The design of the artificial wing is inspired by the curvature of the ladybird’s elytra. It incorporates logic control, motion sensors, power storage, and the active self-righting mechanism within its cavity, utilizing the curvature to encase these elements. Our wing’s geometry is based on a warped semi-sphere along the cockroach’s body, taking cues from the morphology of the Coccinella species. Kinematics also mimics the characteristic oblique movement of the ladybird’s elytra, a unique advantage that allows for a faster recovery compared to other Coleoptera species.

Moreover, artificial wings exhibit superior resilience in contrast to rudimentary constructs like sticks and hinges. Upon freefall, a cyborg insect faces potential damage to either the payload or the insect itself due to the force of impact, this situation is very likely to occur in a disaster situation due to the unstable topology of the environment (Fig. 1). A semi-circular shell diffuses this impact force28, safeguarding the overall integrity of the cyborg insect. In contrast, alternative simplistic structures not only fail to adequately disperse force but may also rebound as a consequence of flexural deformation, resulting in a dual impact. The semi-circular configuration guarantees that the insect can autonomously rectify its orientation across all azimuths.

Fig. 1: Disaster situation scene and overview of the bio-inspired artificial wing attachment on G. Portentosa showcasing wing expansion and contraction and freedom of movement.
figure 1

(Top left) Cyborg insect maneuvering in a demolition area simulating a disaster situation. The insect carries a backpack capable of live-streaming video. During the experiment we found out that the insect tends to fall when approaching edges due to the elevated center of mass. (Top right) Overview of the bio-inspired artificial wing electronic backpack attached to G. Portentosa. In its compact mode (left), the wing enables easy maneuvering in confined spaces. When expanded (right), the artificial limb assists the cyborg insect in performing self-righting motions. (Bottom left) To ensure that the insect’s natural motion is not constrained, a one-point backpack attachment technique is employed. The contact point is positioned on the top of the last head dorsal exoskeleton section. (Bottom right) The backpack-wearing insect demonstrates climbing motion, overcoming our 3D-printed obstacle and angle tool. The demonstration of freedom of movement is completed as the backpack-wearing insect also showcases descending motion.

Full size image

For the initial demonstration of cyborg insect self-righting, a basic artificial wing was employed, featuring a pin connection between the wing and the backpack. This allowed for active rotation of up to 45° relative to the cockroach’s horizontal plane. The rotation was driven by a piston slider crank mechanism, with a miniature slide screw stepper motor (Sxiaoxia U79) serving as a substitute for the piston. The shell’s curvature was designed to facilitate tilting of up to 98° (Fig. 4), which surpassed the theoretical value of 90°, which is the minimum that the insect needs to self-righten in nature26. However, it was observed that in some cases, a 98° angle was insufficient to achieve optimal grip, either due to poor foot contact or to the load excessively lifting its center of mass. It became apparent that a wider tilting angle was necessary, as it would allow for a smoother recovery.

In nature, the expansion of the elytra is driven by the contraction and relaxation of muscles. To emulate this dynamic movement, we transformed the motion from a miniature linear motor to a two-plane oblique motion inspired by the ladybird elytra joint. This resulted in the creation of the bio-inspired artificial wing. The compact mode (Fig. 1) facilitates movement in confined spaces, while the expanded mode (Fig. 1) enables the insect to overcome accidents and prevent the insect from falling on its back in the event of a sudden fall.

The Madagascar hissing cockroach adapts its movement in disaster environments by curling and twisting its body20. To maintain the cockroach’s natural locomotion, the exoskeleton sections have to move freely. The artificial wing is designed with a one-point attachment, limited to the first section of the back (Fig. 1), allowing for a rolling motion both backward and forward. The insect was able to overcome our angle tool obstacle, as seen in (Fig. 1). A 2- or 3-point attachments guarantee a stronger bond between the backpack and the insect, but locking the exoskeleton section may result in preventing the insect from overcoming basic obstacles as observed in the experimental data (Fig. S8).

We compared the self-righting motion of Coccinella with that of G. portentosa, both with and without a load (Fig. 2). The natural self-righting motion of G. portentosa relies on its ability to twist and roll30 into an inverted hyperbolic paraboloid shape (z = x2−y2), resembling a horse saddle (Figs. 2, S1). This geometry creates two contact points at the head and tip of the abdomen, allowing the insect to balance easily at a 90° angle from the ground. By using its legs opposite to the floor, the insect can exert pulling force to achieve self-righting motion. However, the ability to exert pulling force is influenced by the cockroach’s capacity to achieve a consistent grip on the floor. If the floor texture or the tilting position is insufficient for a secure grip, the insect may struggle to self-stand. This challenge becomes more pronounced when an electronic backpack is attached (Fig. 2, S1).

Fig. 2: Self-righting progression of the ladybird insect and G. Portentosa without load, with load and with bio-inspired artificial wing.
figure 2

(Top left) G. portentosa is positioned face-up, and the insect instinctively adopts a saddle stance that enables it to tilt over the side. Once the insect’s body side comes into contact with the ground, it attempts to reach the floor using its opposite legs. After successfully establishing a firm grip, the insect pulls its body weight to complete the self-standing motion. (Top right) When an additional load, such as the electronic backpack, is added, the insect is unable to achieve self-standing motion naturally. (Center) Coccinella is positioned face-up, and the insect’s lack of dorsal mobility, caused by the morphology of its outer elytra, prevents it from generating self-standing motion. Additionally, its shorter legs are unable to reach the floor. Instinctively, the insect resorts to its only available movement: expanding its wings. By exerting a pushing force with its elytra, a back-rolling motion is generated. Once self-standing motion is achieved, the elytra closes. (Bottom) G. portentosa is positioned face-up, with its bio-inspired wing expanded. Similarly to Coccinella, the insect cannot reach the ground with its legs. Thanks to the one-point attachment, the insect can balance itself and generate a rolling motion. Once the abdominal section passes the plane perpendicular to the ground, the rolling motion can be completed. Finally, self-standing motion is successfully achieved. The insect is positioned on top of a transparent surface with controlled inclination using a 3D-printed angle tool.

Full size image

The case of Coccinella is relevant to our study. Ladybirds, when found in a face-up position, are unable to reach the floor with their legs. In such instances, an additional limb is required to assist self-righting27. The expansion motion of the elytra allows the insect to push its body in a unique rolling motion, enabling it to reach the floor (Fig. 2, S1).

Cyborg insect self-righting under disaster situation conditions

Preliminary investigations were conducted to characterize the self-righting motion of the Madagascar hissing cockroach. Initially, the subjects were placed in a face-up position on a flat surface made of 150 gsm paper (Fig. 3) to observe their transition from a supine position to a walking motion. The tests were repeated under falling motion conditions to simulate falling during disaster scenario exploration.

Fig. 3: Showcase frames from the self-standing experiment conducted to assess G. Portentosa’s performance in unpredictable conditions, particularly in disaster areas, while carrying various backpacks.
figure 3

Three adverse studied surfaces are from left to right: flat paper, rocks and soil. (Top left) Bio-inspired wing on paper: success. (Top center) Previous study backpack on stones: failure. (Top right) Optimized backpack on soil: failure. (Bottom left) Success ratios for each of the loads depending on the surface condition. Angle tool is used to determine self-righting success depending on the surface slope. (Bottom center) Bio-inspired artificial wing and (Bottom right) primitive artificial wing tested on 135° slope.

Full size image

To simulate rescue mission scenarios, we replicated two additional types of surfaces that cyborg insects may encounter. The first surface consisted of flat paper (Fig. 3); the second surface consisted of an arrangement of stones ranging from 2 to 5 cm in diameter (Fig. 3); the third surface comprised a layer of hardwood fiber soil (Fig. 3). The insects were dropped from a height of 30 cm onto the aforementioned surfaces. The experiment was repeated for five different conditions: no attachment (baseline), a backpack used in a previous study11, an optimized backpack, a primitive artificial wing, and the bio-inspired artificial wing (Fig. 3).

Reducing the electronic load significantly enhances self-righting, making weight reduction a primary focus in backpack development, as seen in Tables 1 and 3. The curved design of the wing prevented the cockroach from becoming trapped within the rocks or submerged in soil, serving as a protective barrier for the electronics and self-righting assistance. Notably, the bio-inspired wing surpasses all other variants, as shown in the supplementary information (S2).

Table 1 Success rate of self-righting after a free fall (n = 100 per item)
Full size table

Expanded artificial wing performance comparison

The bio-inspired artificial wing exhibited self-righting motion on an angled surface up to a maximum angle of 150°. To characterize the maximum self-righting angle, an acrylic board was attached to a self-designed 3D-printed base with 15° increments. The primitive wing, the bio-inspired wing, and a lightweight design parting from the bio-inspired wing named the “bar-type artificial limb” were dropped on the sloped surface. The bio-inspired wing and bar-type limb demonstrated self-righting motion at a maximum impact angle of 150°. The primitive wing’s tendency was to fall on its side and end up at the 98° angle, not being able to achieve self-righting.

The test was repeated with the backpacks attached to the Madagascar hissing cockroach. The cockroach generated a side-rolling motion while sliding down the acrylic board, utilizing inertia to quickly recover upon reaching the ground (Fig. 3). In some instances, the cockroach may not generate the rolling motion and end up in a tilted position. However, due to the opened wing acting as a supporting structure (Fig. 4), the cockroach can eventually recover. If the cockroach becomes stuck at angles above 150°, it can utilize the bio-inspired artificial wing using the same unique self-righting technique as the ladybird; the cockroach will intuitively balance back and forth, generating a frontal roll movement that culminates in a front flip, enabling it to resume its normal motion (Fig. 2). While the bar-type artificial limb is not capable of generating the ladybird’s roll it outperforms the bio-inspired artificial wing in terms of self-righting time.

Fig. 4: An experiment was conducted to observe the self-standing motion after falling from an angled surface.
figure 4

(Left) Primitive artificial wing, (Middle) bio-inspired artificial wing and (Right) bar-type artificial limb. The results demonstrated a rapid recovery with the bio-inspired implementations. By mimicking Coccinella’s wing expansion, a broader tilting angle can be achieved. Expanding the angle improves the success of self-standing and enhances the overall motion time.

Full size image

The bio-inspired wing is more suitable to self-righten in complex situations as it offers self-righting at a maximum angle of 180°. The bar-type limb is a lighter implementation. The primitive wing tends to reach the 98° angle, which may not be sufficient for self-righting in all cases or may cause longer self-righting times, as shown in Table 2 and videos (S3, S7).

Table 2 Average time needed for self-righting after sliding from the slope (n = 30 per item)
Full size table

Bio-inspired wing mechanical optimization

The development of the wing involved a series of design iterations to attain its final configuration. Drawing inspiration from the ladybird’s distinctive ability to recover, the geometry of the wing was designed to mimic the ladybird’s wing profile. This design was also contingent upon the mechanical tolerances associated with the UV curing resin utilized in the digital light processing (DLP) 3D printing process. The objective of this design was to establish a single, integrated fabrication solution for the on-demand production of a cyborg insect’s backpack. The fundamental design comprises a shell equipped with two pivotal joints that generate an oblique movement inspired by the ladybird elytra joint (Fig. 5). The outer shell serves the dual purpose of connecting both joints and enveloping the cyborg insect’s body.

Fig. 5: Mechanical and dimensional properties of the bio-inspired artificial wing and its bar-type modification.
figure 5

(Top left) Expansion of the bio-inspired wing, the actuator pushes both cranks, the joint points from the wing, and the mechanism can also be observed. For the mechanical simulations, a force is applied in the middle of the wing normal to its surface in order to understand its behavior during a free fall impact. (Top right and center) Using the frontal view, the size comparison between open and closed wings can be seen, and the importance of having a retractable system to reduce the size and allow the insect to pass through gaps can be considered. (Bottom left) Modulation of the motor step time, below 4800 μs, the motor will not increase speed due to missing steps, and above the speed will be decreased as energy would be wasted during stationary periods. (Bottom left and center) Additional design with the purpose of reducing wing total mass alleviating the work of the cyborg insect, on top, we can see the single barred design, and on the bottom dual barred design, the bar connects to the joint points for additional strength.

Full size image

In order to assess the structural integrity and performance under real-world conditions, FEM simulations were conducted in the expanded position. The wing was subject to an impact force normally applied to the curvature of the shell (Fig. 5). The magnitude of the force was computed, factoring in the mass of the cyborg at 17 g (Table 3). The impact force was calculated to simulate a fall from a height of 2 m, equivalent to 3.33 N.

Table 3 Cyborg insect basic electronics and 3D printed wings weights
Full size table

Beginning with the minimal thickness achievable with the UV curing resin, simulations were conducted for shell thicknesses of 0.3, 0.5, and 0.8 mm. The analysis revealed two critical parameters that warrant consideration. The first pertains to the maximum load-bearing capacity, with the most substantial force vector observed at the ball joint. The wing adheres to a lever-like pattern, with the ball joint serving as the pivot and bearing the brunt of the force. The second parameter centers on the material’s flexibility; excessive displacement may result in the bending of the wing to an extent where its functionality is compromised. The 0.3 mm shell showed a disproportional disadvantage with respect to the 0.5 and 0.8 mm shells, as shown in Table 4; this was confirmed by the production of the same and empiric testing. There was not a significant difference between 0.5 and 0.8 mm shells; a 0.5 mm shell was selected as the champion due to its reduced mass. To reduce weight, two paths were taken: one was to add holes in the areas of least mechanical strain (Fig. 5), and the other was to focus on keeping the areas of the wing that were utilized by the cockroach for its recovery (Fig. 5).

Table 4 Bio-inspired artificial wing mechanical simulations
Full size table

In the bar-type design, most of the wing was removed, and only the fall and roll contact areas were kept, the selected thickness for the remaining shell was 0.5 mm (Fig. 5). Due to this, most of the structural integrity was compromised, and the dome-like shape lost much support. To compensate for the loss of areas, an inner rib was introduced along the shape to absorb the energy of the impact, the rib that we call a bar (Fig. 5), which consists of a dual circular sweep along the shape of the wing, the diameter of the circle was varied 0.6, 1 and 2 mm. The resulting part shows worse weight-to-strength performance compared to its predecessor.

The second design path, circular holes were made on the areas of least concern to remove weight, the hole was decided over other geometries due to its similarity to the spots found in the ladybird and its force dissipation. We reduced the weight by 23% while only compromising the structure 8% and increasing its deformation 11% which is equivalent to a 1 mm increase.

In conclusion, the nature-inspired dome-like structure provides improved impact protection against its minimalistic counterpart, while the bar-type structure can offer better maneuvering in confined spaces as seen in Table 5. Deviating from the original ladybird shape requires additional engineering efforts, such as utilizing compound materials or processes to compensate for the loss of structural integrity. Nonetheless it is a path that can be explored in the future with newer multi-material 3D printers, sandwich structures, or coatings.

Table 5 Cyborg insect’s minimum gap size for passing through
Full size table

3D printed mechanism

The designed mechanism for the bio-inspired artificial wing consists of a 1-degree of freedom system with the following components: one screw linear motor, one screw slider, two double rotula cranks, and two one-rotula one-pin cranks for the wings (Fig. 6). The wings are connected to the supporting structure using two 45° oblique symmetrical pins. The supporting structure is solidly attached to the motor.

Fig. 6: 3D printed model with supports, electronic schematic and microcontroller program.
figure 6

(Top left) 3D printing models of the bio-inspired artificial wing backpack are displayed. Support columns, indicated in red, can be manually removed after the curing process. Support beams, highlighted in green, require the use of a cutting tool for removal after the curing process. (Top right) The circuit diagram illustrates the active self-standing assistance system hardware components. (Bottom) Workflow for the active self-standing mechanism loop. To switch from shock detection to angle detection, the block marked in red can be excluded to disregard acceleration.

Full size image

With the current dimensions, the mechanism converts a 5 mm slider travel into a 45° oblique rotation, extending the tilting angle of the wing to 112°, enabling the cockroach to easily reach the ground plane. The motion can be observed in the supplementary information (S5, S6).

The rotula joints are finely adjusted to create a snapping fit with a 6-degree-of-freedom movement, taking into account the tolerance of the 3D printer. All 3D printed parts are designed to be printed parallel to the layering of the printer to ensure dimensional stability. Additional supporting structures have been included to facilitate easy removal (Fig. 6).

The design offers versatility, allowing for the adjustment of the torque to total angle displacement ratio by modifying the positioning of the wing rotula connection and altering the crank length. The mechanism functions as a piston slider crank with a variation in the rotation axis, which is oblique instead of normal to the system, resulting in wing expansion (Fig. 5). The crank length used for the subsequent experiments was determined based on the performance of the active system, maximizing the displacement for the torque produced by the battery-powered driver. The assembly of the system can be observed in supplementary information (S4).

Insect adaptation to a new attachment

An observational study spanning one month was conducted subsequent to the attachment of a backpack to a cockroach in order to comprehensively gauge its impact on the insect’s behavior. This study differs from other behavioral studies which primarily account for obtaining food, a basic necessity31,32. Most notably, the insect exhibited a discernible likeness to evade intense light sources and seek refuge in the shelter of shadows. Preliminary trials revealed that the cockroach employs a trial-and-error decision-making process. When confronted with an orifice, the cockroach’s initial instinct is to attempt to maneuver into it with the objective of accessing a shadowed refuge. The duration of these attempts varies, but once abandoned, the insect systematically seeks alternative entry points along the surrounding perimeter (S9). This behavior is beneficial in exploration activities, the insect can evaluate whether a goal is possible to be reached, saving stamina and seeking alternative routes.

Furthermore, the study examined the cockroach’s ability to navigate through confined spaces while wearing the backpack. It was observed that with the backpack, the insect can contort itself to a height of 17 mm with a width of 25 mm. This gap accommodation is made possible due to the flexibility of the 3D-printed material. However, when electronics are integrated into the backpack, the height requirement increases to 20.5 mm owing to the additional rigidity imparted by the electronic components, which coincides with the theoretical limit of the design of the backpack (Fig. 5). The steering of the cyborg insect should consider gap sizes and employ vision technology, such as camera imaging or radar sensing, to enhance navigational precision and adaptability.

Actuator characterization

The selection of a micro stepper motor for our application stems from the necessity for a cyclical actuator exhibiting robust passive position fixation and torque capabilities sufficient to propel the artificial wing in an oblique motion. In this study, the Sxiaoxia U79 model was chosen as the primary actuator, given its compact construction and integrated infinite screw. While MEMS or piezo electric actuators offer heightened precision in positioning owing to their reduced step size, this precision comes at the expense of speed and increased power consumption. Given the absence of precision requirements in our application, opting for a stepper motor is deemed the most optimal choice.

The motor underwent comprehensive characterization procedures aimed at determining its optimal operational parameters. Tests were conducted to evaluate speed, power consumption, maximum load-bearing capacity, and initial motion conditions. The motor’s maximum load-bearing capacity is 6.1 g when the infinite screw is parallel to the ground and 6.2 g in the vertical position (Table 6). In simulating real-world scenarios involving the artificial wing carrying a load, speed tests were systematically conducted at load levels of 0.6, 1, and 1.4. Through these assessments, the optimal step duration was determined to be 4800 μs (Fig. 5). It was observed that a shorter duration resulted in missed steps, while longer durations negatively impacted speed and power consumption.

Table 6 Actuator maximum load
Full size table

The system was subjected to cyclical operation tests to establish its maximum repetition capacity. A full use cycle of the artificial wing takes 3.6 s consuming 81 mA, for this purpose a LiPo battery with a discharge of 2 C is required. If it were to be used continuously with a 40 mAh battery, the system could theoretically achieve a total of 493 repetitions or 29 min. The duty cycle employed in these experiments involved the concurrent utilization of the gyroscope sensor, contributing to the system’s overall performance assessment.

Our system employs two batteries in series accounting for 8.2 V, a power converter is used to transform the potential to 5 V needed to drive the actuator and the main control unit, this creates energy waste compared to an ideal power supply. Nonetheless, as two batteries are being used at the same time the power load is divided (Table 7). An additional implementation utilizing 3 V electronics was developed, but the actuator’s speed and torque were not sufficient for the purpose of this research.

Table 7 System power consumption
Full size table

Active self-righting assistance

In the initial experiment the insect was laid on its back with the closed bio-inspired artificial wing, then the expanding motion was activated. The force of the motor was not enough to lift the insect in a timely manner, causing a waste of battery energy and insect stamina.

Expanding the wing before the insect touches the floor prevents energy waste and speeds up the process of self-righting. The stepper motor is used to lift only the weight of the wing, while the linear screw passively constrains displacement once the movement is finished. To monitor the orientation of the cyborg insect, an additional gyroscope and accelerometer sensor (MPU-6050 6-axis MEMS motion tracking) were integrated into the main microcontroller unit (Arduino Pro 5 V 16 MHz) (Fig. 6).

Two techniques have been implemented to activate the expansion of the wing: shock detection and angle detection. The wing expansion process takes a total of 0.9 s. In the case of shock detection, assuming an ideal scenario where the accident is detected instantly, the wings would require a minimum free fall distance of 3.97 m to fully open. For angle detection, a specific angle threshold is defined to activate wing extension (Fig. 6). The threshold is set to activate the wing during activities such as climb or descending, where the cockroach has a higher chance of falling.

The shock detection system (S5) functions by activating the wing expansion when the cyborg insect’s backpack is shaken under an angle that could cause it to lose grip and fall on its back. The wing expands to protect the insect from becoming stuck, and once the angle is recovered, the wings return to their initial position, compacting the backpack to ensure the insect can move through confined spaces again. However, this technique may face challenges if the falling distance is shorter than the total time required to expand the wings, as the movement may not be completed before contacting the ground plane.

The angle detection system (S6) uses the same angle consideration to determine if the cockroach is in a situation that could potentially cause it to become stuck. Once the determined angle is reached, the wings remain open until the insect is back on the ground plane. This technique guarantees that the wings are open in time to prevent a collision with the ground in the event of a fall. A trade-off with the shock detection technique is that the actuator element is used more frequently in angle detection, as it does not take into consideration the actual need for self-righting assistance by the cyborg insect.

In the field experiment (Fig. 7 and S6) using the single-point backpack attachment method on G. portentosa, the configuration setting was angle detection. The bio-inspired artificial wing successfully expanded after detecting the cockroach’s initial climbing. The cockroach was then pushed to simulate an accident and evaluate the effectiveness of the backpack in assisting self-righting. As expected, the additional tilt angle provided by the wing allowed for a longer reach of the cockroach’s legs and successful self-righting. Once the cyborg insect resumed its walking motion, the wings contracted for more agile ground displacement. The process was tested multiple times, demonstrating the repeatability of the mechanism (Fig. 8).

Fig. 7: Experimental results of the active self-righting mechanism.
figure 7

(Top left) An experiment on angle detection for active self-righting assistance is conducted. The cyborg insect is positioned near a tree trunk to initiate climbing motion with its wings in compact mode. (Top center) As the climbing motion begins, the wings transition to the expanded mode to avoid getting stuck in case of an accident. (Top right) Manual induction of falling is carried out. (Bottom left) The added tilting angle enables the cyborg insect to reach the ground with all of its legs. (Bottom center) Walking motion is achieved. (Bottom right) The wings return to the compact mode to avoid obstacle collision, the process can be repeated in a loop.

Full size image
Fig. 8: Evaluation of the performance of the machine learning model to operate the activation of the artificial wing mechanism and its use example.
figure 8

(Top left) Classification report of the trained random forest machine learning model, we can observe a discrepancy in the falling motion status. (Top right and center) The discrepancy is confirmed once test data is analyzed by the model. Its lower correctness may be caused due to the smooth progression of falling motion in the Z-axis, which is difficult to identify unless velocity is high. (Bottom left) Sequence of cyborg’s tree climbing followed by recovery and fall, this action is repeated several times in a test run. (Bottom right) We can observe the easiest and most difficult status to identify by our model, while “Plane stationary” has a 90.5% success ratio and may sometimes be mistaken as “Plane motion”, (Bottom left) falling down has only a 25.8% defect ratio and can be mistaken for any of the other statuses, the implementation of an additional sensor such as ground contact would provide a valuable dimension to the model.

Full size image

Regarding mechanical stability, it was found that the wing could withstand impacts and scratches during practical use. However, the rotula connection sometimes detached depending on the angle of impact. This is due to the fine adjustment required to allow for snap assembly while guaranteeing the three degrees of freedom of the bal-and-socket joint. This design decision was made to reduce weight and mechanism complexity by utilizing a single part for snap assembly.

Machine learning classifier to optimize wing activation

To expedite wing activation response, a machine learning classifier system was developed to identify the movement status of the cyborg insect. We selected a random forest machine learning classifier due to its capacity to manage high-dimensional data without necessitating dimensionality reduction or feature selection and the possibility of providing a prompt classification33,34,35,36.

We collected information from the gyroscope while the cyborg performed climbing and falling motions. The time areas were manually classified to differentiate motion patterns—walking, climbing, stationary, and falling. Upon detecting a falling status, the model triggers an activation signal for wing expansion and inversely retracts the wings upon recognizing a return to walking motion.

The results showed that the most difficult status to identify is the falling motion (Fig. 8). A falling acceleration may not be easily distinguished until midfall, this may be causing the classifier to produce less precise status identification. In the future, this will be solved by adding additional dimensions to the model with the use of additional sensors, such as a contact sensor between the body and the floor.

In order to implement the classifier into our current system, we compressed the model and reduced its size to 986 kb. However, our microcontroller does not have sufficient memory to operate the model and relies on wireless communication to outsource the data processing, a challenge towards real-time responsiveness.

In forthcoming iterations, the integration of self-decision-making optimized chips to independently effectuate responsive actions is poised to address these limitations37,38, ultimately paving the way for the realization of self-guided robotics for deployment in disaster-stricken areas.

Conclusions

The research conducted on cyborg insects has provided valuable insights into enhancing their self-righting capabilities in disaster situations. By studying and applying biological solutions observed in other organisms, such as the ladybird, the researchers were able to extend the mobility of cyborg insects. This highlights the importance of drawing inspiration from different life forms to overcome mobility limitations.

DLP 3D printing has proven to be a powerful tool for fabricating miniaturized mechanisms, which is advantageous in disaster scenarios where resources and time are limited. The ability to produce artificial appendages on-site enables the customization and adaptation of cyborg insects for specific rescue missions, enhancing their overall mobility and effectiveness.

Machine learning (ML) will also play a significant role in improving the decision-making process for activating the artificial wing. By implementing real-time classifiers, the activation of the wing can be optimized to conserve battery life and enable longer rescue missions. This intelligent system will enhance the overall efficiency and autonomy of cyborg insects during operations.

The use of a repeatable actuator ensures continuous functionality and reliability in overcoming obstacles during lengthy and unpredictable rescue missions. This reliability is crucial for applications where quick responses and the ability to access confined spaces are essential for successful rescue operations. The research conducted on cyborg insects and their enhanced mobility sets a foundation for further advancements and the potential utilization of these organisms in critical scenarios.

Methods

Experimental animals and breeding environment

The breeding and maintenance of male and female adult G. portentosa (Madagascar hissing cockroach) involved creating controlled terrarium environments. The soil in the terrariums consisted of a mixture of wood fibers and barb fungus, specifically the Marukan Insect Mat. The temperature inside the terrariums was regulated using a heating mat, maintaining a constant temperature of 30 °C.

To ensure proper nutrition for the cockroaches, they were provided with glucose cups. These cups, known as Fujikon Wide Cup Forest Tree Sap, served as a food source for the cockroaches.

In order to conduct further studies, specimens with higher activity patterns were identified and separated from the breeding group. These selected cockroaches were placed in individual containers to facilitate focused observation and experimentation. They were provided with the same environmental conditions and care as the breeding group, ensuring consistency in their living conditions.

Invertebrates, including cockroaches, do not require ethics approval for animal research according to the National Advisory Committee for Laboratory Animal Research.

Self-righting success from face-up position

We used a 1080p 60fps generic camera to observe the transition of G. portentosa from a face-up position to walking motion on top of a 150 gsm white paper.

We repeated the experiment while attaching a backpack with basic elements for cyborg control, as referenced in a previous study15. We also attached our primitive artificial wing to the specimens of G. portentosa. The same experiment was conducted using specimens of Coccinella septempunctata, the seven-spotted ladybug.

Self-righting success after a free fall

We conducted drop tests on G. portentosa from a height of 30 cm, examining five different load conditions and three different environmental conditions. The load conditions were as follows: No load (control); Backpack design from a previous study15; Our backpack design with optimized components; Our primitive artificial wing; Bio-inspired expanded artificial wing.

The terrain conditions for the tests were: First surface: 150 gsm white paper; Second surface (A): Plant soil composed of wood and vegetal fibers with a thickness of less than 1 cm; Third surface (B): Stones ranging in size between 5 to 10 cm in diameter.

The entire motion during the drops was recorded using a generic camera.

Self-righting success at an angle

To conduct the experiment, a specially designed 3D printed angle control tool was created to securely hold a transparent acrylic board measuring 150 × 150 mm. The tool allowed for adjusting the board in increments of 15°.

Two types of backpacks were tested: a primitive artificial wing and a bio-inspired artificial wing equipped with a microcontroller and a single battery unit. The backpacks were positioned parallel to the board, making contact with the outer shell of the artificial wing. The setup was then allowed to slide down the board. This process was repeated for each angle ranging from 90° to 150°.

The experiment was performed again, but this time, the backpacks were mounted on the G. portentosa using the one-point attachment technique. The results of the experiments were recorded using a generic camera for further analysis.

Design of bio-inspired artificial wing

In order to achieve accurate wing design for the artificial system, frontal and side views of Coccinella septempunctata were subjected to projection on CAD drawing software. Subsequently, the design underwent a one-dimensional warping process to conform to the dimensions of G. portentosa. The resulting volume was then hollowed out using a 0.5 mm shell function.

The electronic components and actuator were meticulously modeled and positioned within the cavity of the wing structure. The artificial wing, comprising seven individual parts, including two wings, two cranks, a base, a motor-to-wing connector, and a pin cover, was subjected to kinematic simulations to determine the maximum wing span. Additionally, adjustments to the crank connection points and overall length were made to regulate the maximum wing extension.

Precise tolerances were carefully considered, accounting for the fabrication parameters of the DLP 3D printer and the designated print direction. A tolerance of 0.2 mm in the spherical radius was implemented for the rotula connection, while a tolerance of 0.1 mm was set for the pin connection in the cylinder radius.

In order to reduce weight and optimize the center of gravity, the bio-inspired wings were designed with strategically placed holes. These holes served the purpose of preventing the insect from flipping upside down when dropped. The choice of circular holes was inspired by the distinctive spots found on the Coccinella septempunctata species.

To ensure dimensional stability and facilitate the assembly process, a thoughtful design approach was employed, incorporating support structures. The contact surface of these structures featured an extrusion with a diameter of 1.5 mm and a height of 0.3 mm, followed by a subsequent extrusion of 0.5 mm in diameter, delimited by the contact of the final part. These structures were specifically engineered for ease of removal without the need for additional tools. Additionally, an additional support structure was strategically integrated into the motor-to-wing connector, aiming to enhance dimensional stability during the curing process, with the intention of later removing it using a cutting tool.

By implementing these design considerations and methodologies, the dimensional accuracy, structural integrity, and ease of assembly were significantly improved, contributing to the successful development of the artificial wing system.

Fabrication of bio-inspired artificial wing

The fabricated parts were produced utilizing an Anycubic Photon D2 DLP 3D printer, employing Siraya Tech fast curing Resin in Navy blue color. The printing parameters were configured with an 8-s normal exposure time, 1-s off time, 40-s bottom exposure time, 6 bottom layers, and a layer thickness of 0.1 mm. All other parameters were maintained at their default settings. Notably, no support structures were incorporated from the slicer software, and the parts were directly positioned on the building surface.

The entire printing process took approximately 1 h and 10 min. Once the printing was completed, the parts were carefully removed and subsequently immersed in an ethanol bath containing Sanis Alcohol 75. The parts were subjected to stirring using an ultrasonic cleaner (Onezili OZL-800) operating at a frequency of 40 kHz for a duration of 5 min. Afterward, the parts were allowed to dry before being transferred to a UV curing box (ELEGOO Mercury X) for an additional 5 min of exposure.

For the assembly of the wings onto the base, a small amount of superglue was utilized to securely seal the pin cover. Similarly, the motor, situated within the base slot, was affixed using a single-point application of superglue. Care was taken to avoid any inadvertent spillage of the adhesive onto the screw while gluing the motor to the wing connector. The crank connections were assembled through a snapping mechanism, ensuring a secure attachment. Furthermore, the electronic board was fitted into the base holder, completing the assembly process.

The entire device assembly and subsequent mechanism testing procedures were documented using a generic camera capable of capturing footage at 1080p resolution and 60 frames per second. Additionally, a smartphone (Samsung A53 5 G) was employed to record relevant visual information during the testing phase.

Attachment of bio-inspired artificial wing

The Madagascar hissing cockroach overcomes obstacles (Fig. 1) and performs self-correct (Fig. 2) by altering its body shape. However, even in the most recent studies11,12, there remains a preference for affixing the backpack to the insect’s dorsal surface to ensure stability during level-ground ambulation, thereby constraining the cyborg insect from altering its body shape. Therefore, backpack assembly in this study was attached to the insect by the one-point attachment base onto the first section of the dorsal area using thermal glue.

The insect was subjected to a 15-second exposure of CO2 to induce sedation. Subsequently, thermal glue was applied to a 3D-printed base and allowed to cool until reaching a lower temperature. Once the thermal glue had cooled, it was gently pressed against the dorsal section of the insect for the purpose of reshaping. Following experimentation, the attachment could be easily removed by applying a firm pulling force.

One, two, and three-point attachment of bio-inspired artificial wing

In order to assess the efficacy of employing the one-point attachment method, a specific experiment was conducted, necessitating the insect to execute rolling movements for maneuvering. The experiment involved equipping the insect with a lightweight wood plank, roughly matching its body footprint. Subsequently, the insect was placed on a 150 mm diameter wood log. Given the elevated ambient illumination, prompting the insect to seek shadows, it descended the wood log. The experiment was iterated, altering the setup by attaching the wood plank attaching to one, two, and three exoskeleton head sections. This iterative process aimed to discern the impact of varying attachment configurations on the insect’s maneuverability.

Active self-righting circuit

The active self-righting system incorporates a circuit comprising key components, including a microcontroller unit (MCU) in the form of Arduino Pro, operating at 5 V and 16 MHz. The system further integrates an accelerometer and gyroscope module, specifically the MPU-6050 6-axis MEMS motion tracking device. Additionally, a miniature slide screw stepper motor, designated as the Sxiaoxia U79, is integrated into the circuit. To power these components, two 3.7 V LiPo batteries with a capacity of 40mAh each are utilized.

To ensure the appropriate voltage level for driving the motor, the two 3.7 V LiPo batteries are connected in series. This configuration allows the combined voltage of 7.4 V to be supplied to the Arduino Pro voltage converter, which in turn powers the built-in AtMEGA328P microcontroller responsible for driving the motor. Moreover, the MPU-6050 module is connected to the Arduino Pro power supply and establishes communication with the microcontroller using the I2C protocol, facilitating data exchange and control.

By integrating these components and establishing the necessary power and communication connections, the active self-righting system can effectively carry out its intended functions and perform coordinated movements based on input from the accelerometer, gyroscope, and motor control.

Active self-righting assistance program

The dynamic behavior of the bio-inspired wing was implemented through programming using the Arduino Integrated Development Environment (IDE). The corresponding code was then uploaded to the microcontroller unit (MCU) using an FTDI generic board. The code encompasses two distinct self-righting algorithms, namely shock detection and angle detection. It should be noted that these modes cannot be executed simultaneously, as the desired mode needs to be selected by modifying the “MODE” variable before the code is flashed onto the MCU.

Upon initialization, the program establishes the ground plane by capturing the initial position data from the MPU (accelerometer and gyroscope module). This information serves as a reference for subsequent operations. In the event that recalibration is required, the reset button integrated into the Arduino Pro can be pressed to facilitate the recalibration process.

Component weight

Milligram scale (LACHOI 200 g × 0.0001 g) was used to measure 3D-printed parts, hardware for active system control, and experimental subjects.

Cockroach maximum load and endurance

The maximum load capacity of the one-point attachment method was determined by conducting an experiment with G. portentosa. A specially designed 3D-printed structure was affixed to the insect, allowing it to move freely within a confined space. To stimulate walking, gentle contact was applied, and the insect’s behavior was observed.

The results indicated a clear relationship between the load in the backpack and the insect’s locomotion. As the load increased, there was an observable trend of the insect exhibiting a greater likelihood of stopping walking. Notably, a load of 15 g was found to be the maximum weight at which the insect consistently came to a complete stop.

Artificial wing self-righting active test and experiment

In order to evaluate the performance of the self-righting shock detection and angle detection systems, the helmet component was appropriately connected to the FTDI generic board, which facilitated both the acquisition of sensor data and the provision of an external power supply. The collected data served as the basis for determining the threshold values required to activate the expansion and retraction functionalities of the bio-inspired artificial wing.

To assess the efficacy of the shock detection system, the experimental setup involved linking the device to a portable power supply comprising two series-connected 3.7 v LiPo 40 mAh batteries. The device was subjected to controlled shaking at various positions, enabling an assessment of its shock detection capabilities. These experiments were meticulously captured using a generic camera to facilitate subsequent analysis.

Regarding the angle detection system, the experimental investigation entailed the attachment of the bio-inspired artificial wing to G. portentosa utilizing the one-point attachment method. Employing the same portable power supply as aforementioned, the angle detection system was subjected to rigorous testing. Specifically, the cyborg insect was positioned on a tree bark to initiate the climbing motion, accompanied by the expansion of the backpack wing. Subsequently, the manual force was applied to the insect, simulating an accident scenario within a disaster context. The resulting motion was recorded utilizing a Samsung A53 5G smartphone, allowing for detailed analysis and evaluation.

Actuator characterization

The motor was characterized by using the main control unit (ATmega328p) as a driver. For the speed tests, loads 0.6, 1, and 1.4 g were attached to the screw connector, the amount of time supplied to the coils was varied in steps of 200 μs and the initial position and direction of the movement were considered, parallel motion, back and forth. To determine the maximum load the same screw connector attachment was used, to consider the success of the motion the motor must complete the whole trajectory without miss steps, the weight was increased until the miss step was found. For the determination of the power consumption a power supply with an amperage control was connected instead of batteries, through the inbuilt amperemeter current was determined, 5 V tension was used to consider optimal conditions while 8.2 V was used to compare with battery usage.

Minimum gap size pass characterization

In order to ascertain the cockroach’s threshold for traversing confined spaces, a rectangular aperture measuring 25 mm by 25 mm was incised into the wall of an enclosed container. This experimental setup was illuminated by a white LED light source. As the cockroach actively sought refuge from the radiant light, its attempts to enter the box through the prepared opening were documented, classifying each endeavor as either successful or unsuccessful. The dimensions of the aperture, namely the width and height, were progressively reduced in a stepwise manner to delineate the minimal gap required to enable successful passage by the cockroach.

Backpack awareness and behavior

The experiment involved the presentation of the cockroach with a choice between two gaps within a confined enclosure: one gap characterized by a smaller spatial dimension rendering passage impossible and another gap distinguished by its adequate size for easy traversal. Subsequently, the cockroach was afforded the liberty to walk freely within the enclosure and make an independent decision concerning which of the two openings to attempt passage through. The observed actions of the cockroach were recorded and classified into distinct categories: “ceased movement”, “circumnavigated the box”, “located the second hole and successfully entered”. This experiment served as a means to gain insights into the cockroach’s cognitive awareness of its own physical constraints and its capacity to make informed decisions based on its perception of the environment.

Machine learning algorithm implementation

For the training of the machine learning model, a tailored backpack for data collection was placed on the back of one specimen of Madagascar cockroach. The data was logged into a computer using a Wi-Fi module (NRF24L) for data transmission. The cockroach was let to walk, climb, rest and fall, the collected data was labeled accordingly. For the development of the software, the scikit-learn Python module was used.