A Soft Growing Robotic Endoscope for Painless and Strain-Free Insertion

Abstract

Numerous studies have attempted to develop medical devices using vine robots due to their potential for frictionless locomotion and adaptability in confined environments. However, for applications in colonoscopy, challenges such as high stiffness, limited steering capabilities, difficulties in integrating tethered sensors, and issues related to safe retraction have hindered their practical application. This article addresses these challenges and presents a comprehensive solution that simultaneously resolves these issues while preserving the intrinsic features of vine robots. We propose a novel soft robotic endoscope that leverages an optimized eversion mechanism to maintain low stiffness and ensure compliance with the natural curvature of the colon, minimizing bowel distension. To enable real-time imaging, we introduce a passive tethered camera stabilization system that secures the camera at the distal tip with minimal internal tension. Additionally, the device integrates active steering capabilities using fabric pneumatic artificial muscles, allowing for precise two-degree-of-freedom steering to navigate through complex pathways. A non-sealed, self-retractable mechanism ensures safe and reliable retraction by preventing buckling while maintaining the robot’s compliance, even with an embedded tethered sensor inside the inner channel. Comprehensive characterization of key parameters, such as vine diameter and retraction channel geometry, further enhances the system’s performance in endoscopic applications. The effectiveness of the proposed endoscope was validated through extensive testing in endoscopic phantom models and in vivo trials, demonstrating significant reductions in insertion forces and colon deformation compared with conventional endoscopes. In phantom studies, the device demonstrated an 80% reduction in mesentery extension compared with a conventional flexible endoscope. In vivo, the soft growing endoscope (SGE) reached the ileocecal valve within 2 min while maintaining real-time imaging, internal channel integrity, and buckling-free retraction. By overcoming key challenges in adapting vine robots for endoscopy, this SGE offers a minimally invasive, safer, and more effective solution for colonoscopy, enhancing patient comfort and procedural efficiency while reducing physical strain on physicians.

Keywords

vine robots soft growing robots surgical robotics robotic endoscopy

Introduction

Colonoscopy is an essential procedure for the diagnosis and treatment of intestinal diseases. Flexible endoscopes (FEs) have been the gold standard device for colonoscopy. These FEs feature a camera at the distal end, along with a semirigid proximal bending section that can be steered.¹ The procedural technique involves manually advancing the endoscope to the cecum by leveraging the physician’s dexterity in manipulating the device.² During this process, the endoscope navigates through the intestine by sliding along its walls, with air insufflation used to expand the lumen, facilitating exploration.

However, this manual insertion exerts pressure on the intestinal walls and distension of the colon, leading to several procedural problems. First, because of bowel distension during colonoscopy, many patients experience abdominal pain.³ For this reason, colonoscopies are usually performed under sedation, which introduces the additional risk of sedation-related complications. Second, bowel perforation can occur due to injury to the bowel mucosa from the tip of the endoscope during advancement.⁴ Third, unlike upper endoscopy, colonoscopy requires frequent use of techniques such as pushing and torquing the shaft to advance the endoscope to the cecum and more manipulation of the tip using the control levers. These maneuvers place significant physical strain on the body, potentially leading to musculoskeletal disorders in endoscopists over time.⁵

Recently, in the field of soft robotics, an innovative robot capable of efficiently navigating narrow environments has been developed. Known as the vine robot, it utilizes eversion actuation mechanisms, enabling frictionless locomotion and high adaptability in complex environments.^6–9 Vine robots have been applied in fields such as search-and-rescue operations and inspection tasks. Additionally, their potential has been explored in medical applications, including targeted catheter delivery and minimally invasive surgery.^10–13 Notably, the eversion mechanism allows the robot to advance without significant frictional resistance, a critical feature for navigating the long and tortuous pathways of organs.¹⁴

Despite these promising features, significant challenges remain when adapting vine robots for use in highly flexible and soft environments like the colon. The stiffness induced by inflation can disrupt the surrounding tissue structure, while the integration of tethered sensors, such as cameras, increases mechanical complexity and internal tension, making miniaturization and operation more challenging. Additionally, retraction during operation may lead to buckling, which poses a risk of tissue damage. To address these challenges, a practical endoscopic vine robot must satisfy several key requirements:

Retain the soft, morphing characteristics of vine robots while maintaining low stiffness.

Enable the integration of tethered sensors, such as cameras, without compromising the robot’s performance.

Preserve the frictionless eversion mechanism for navigating the tortuous pathways of the colon.

Incorporate steering capabilities to traverse convoluted and narrow environments.

Ensure safe retraction to prevent damage to surrounding tissues.

Various studies have attempted to tackle these issues individually, such as attaching cameras to the robot’s tip,^15–16 reducing buckling during retraction,^17–21 and incorporating steering capabilities.^22–27 However, achieving an integrated solution that addresses all these requirements simultaneously remains a significant challenge, particularly in the context of the soft and narrow environments encountered during colonoscopy.

In parallel with these efforts, several soft robotic growing endoscope platforms have been proposed to enhance compliance and reduce insertion forces during colonoscopy. These systems employed diverse actuation and navigation strategies, contributing valuable insights into endoluminal locomotion.^28–33 However, many were evaluated in simplified or rigid phantoms that did not capture the anatomical complexity of the colon and often lacked integration of key functions such as retraction, tethered sensor integration, and effective tip steering. As a result, despite their promise, these platforms remain limited in demonstrating practical applicability in anatomically relevant and clinically realistic settings.

In contrast to previous studies that addressed individual challenges in isolation, this work achieves functional integration of all essential capabilities, including growth, retraction, tethered imaging, and active steering, within a compliant soft growing robot. This integration enables a complete in vivo colonoscopy while maintaining real-time imaging and preserving internal structural integrity. In this article, we present several technical breakthroughs that address the aforementioned challenges, resulting in an integrated solution that satisfies all key requirements for endoscopic procedures without compromising the function or performance of any individual capability. These breakthroughs include:

Soft and low-stiffness eversion mechanism: By optimizing the vine’s parameters through comprehensive characterization and maintaining the simplest design of the eversion mechanism, operational pressure and internal tension remain low. Consequently, the proposed eversion mechanism ensures low stiffness while preserving the soft, morphing characteristics of vine robots. This innovation enables the robot to conform to the colon’s natural curvature without causing significant bowel distension.

Integration of a passive tethered camera stabilization system: A novel passive synchronization mechanism secures the camera’s position at the robot’s distal tip while minimizing internal tension and friction. This system ensures consistent real-time imaging during both navigation and retraction while preserving the low stiffness of the robot.

Frictionless locomotion for tortuous pathways: By fully leveraging the eversion mechanism with minimal design modifications, the robot achieves frictionless tip-growing locomotion with low pressure and minimal resistive forces, enabling navigation through tortuous and mucosal colon pathways without the need for air insufflation. This reduces patient discomfort and procedural complexity.

Active and passive steering capabilities: The robot incorporates fabric pneumatic artificial muscles (fPAMs) for two-degree-of-freedom (2DoF) active steering while also relying on passive adaptability to the colon’s natural structure. The proposed method’s passive steering is highly reliable due to the robot’s low operational pressure and bending stiffness, allowing it to naturally conform to the colon’s shape during insertion and retraction. This combination of active and passive steering further enhances navigation through sharp turns and convoluted environments.

Non-sealed self-retractable (SR) mechanism: A unique retraction system was introduced to prevent buckling during withdrawal, ensuring fast, safe, and reliable retraction. This mechanism preserves the device’s compliance and simplicity. Furthermore, the tethered sensor can be embedded within the vine’s inner channel during the retraction process, which is essential for practical endoscopic applications.

These technical advancements enable the practical application of a vine robot for endoscopic procedures and pave the way for safer, more effective, and patient-friendly colonoscopy techniques.

Materials and Methods

Figure 1 illustrates the advantages of utilizing the eversion mechanism for endoscopy. The frictionless growth characteristic of the eversion mechanism eliminates the need for air insufflation during the insertion process, and its inherent compliance results in minimal distension, thereby reducing the risk of perforation and effectively addressing issues related to patient discomfort during endoscopy.

FIG. 1.

Advantages of the soft growing endoscopy system. The system enables wired-camera-based intestinal diagnostics, providing real-time imaging during endoscopic procedures. Its low stiffness allows it to conform to the shape of the colon, while the soft morphing nature ensures robustness against spasms. The frictionless growth mechanism enables smooth operation even in sticky intestinal environments. Designed with low-cost, disposable materials, the system helps prevent infections and is suitable for single-use applications.

The essential requirements for adapting vine robots to endoscopic applications are closely related to the fundamental functionalities that an endoscope must perform. An endoscope must be easily insertable, capable of accommodating tethered sensors, and safely retractable. Among the factors to be considered in achieving these features, the most critical is the environment in which the vine robot interacts—an environment that is convoluted and soft. Therefore, it is crucial to design a system that preserves the inherent compliance and adaptability of the vine robot. This section details the design approaches to achieve each of these characteristics. Figure 2A illustrates the hardware components of the developed soft growing endoscope (SGE). Each feature introduced in the SGE operates independently, ensuring that one function does not interfere with another.

FIG. 2.

Hardware configuration of the soft growing endoscope. (A) Overview of the key hardware components, including a detailed view of the transporter mechanism inside the base and the configuration of the distal tip. (B) Image of the everting membrane, with cross-sectional illustrations showing the shaded areas where pressure is applied during eversion and retraction. (C) Right-side view of the base, highlighting the positioning of the antagonistic motors and the arrangement of the everting membrane.

Eversion-based propulsion

The primary mode of propulsion for the SGE is based on the eversion mechanism. Figure 2B shows the eversion mechanism of the SGE. In this mechanism, a thin membrane everts and extends forward when pressurized, with the inner membrane continuously everting to form the outer shell as the endoscope propels forward (Fig. 3A). This eversion-based propulsion allows the SGE to navigate the intestinal walls with minimal friction, adapting to the shape of the colon and thereby reducing the need for air insufflation during insertion, as well as stress and deformation on the intestinal tissues.

FIG. 3.

Schematic of pressure flow. (A) Growth. Pressure is applied to the main body via the base, while the transporter moves forward to drive eversion. Pressure is passively synchronized with the inner channel containing the tether through an open junction. Steering can be performed simultaneously. (B) Steering. Actuation is achieved by pressurizing one or more of the three circumferential steering channels (blue arrows). A solenoid valve traps pressure, and the pressurized main body prevents backflow. Steering can occur concurrently with growth. (C) Retraction. Air flows from the tail into the retraction (steering) channels (blue arrows) and exits through the outlets at the proximal end (red arrows). The transporter pulls the tail, while a valve regulates exhaust and steering pressure. (D) Airflow during retraction. During retraction, high-velocity air is introduced through the retraction channels from the base and exhausted through outlets located at the proximal end of the steering channels. The airflow generates tip-driven inversion by fluid momentum. A sharp 180° turn and orifice-like constriction at the tip create a local pressure drop, reducing the force required for inversion. Meanwhile, the outer membrane is intentionally depressurized, which lowers its stiffness. As a result, the whole structure behaves like an inflated cylinder (inner membrane) simply being pulled back, enabling fundamentally buckling-free retraction without compromising compliance.

The SGE’s eversion mechanism comprises a cylindrical everting section that forms the outer shell as it propels forward, and a non-everting section that remains inside, consisting of three retraction channels extending from the steering actuators attached to the everting section. This design facilitates the integration of the tethered camera, steering, and retraction without mutual interference. Cross-sectional illustration in Figure 2B shows the shaded areas where pressure is applied during the eversion and retraction phases.

The proposed eversion mechanism was meticulously designed to closely maintain the original geometry of the vine robot. Because no additional hardware components interact with the membrane, the robot can grow while maintaining low pressure. This, in turn, results in low bending stiffness of the robot. Consequently, the proposed mechanism allows the robot to conform to the winding structure of the intestine and grow with ease while minimizing deformation of the intestinal walls.

Embedded tethered camera

A vital component of the endoscope is its imaging system, which plays a crucial role in providing visual feedback during procedures. However, the eversion mechanism involves relative motion between the membrane and its tip, necessitating additional mechanisms to maintain the imaging sensor at the tip.¹⁵ Furthermore, to effectively utilize a tethered sensor within the eversion mechanism, it is essential to mitigate the friction caused by the pressure exerted on the inner membrane.

In this study, we implemented a passive synchronization mechanism that stabilizes the internal membrane pressure to secure the working channel and maintain the position of the tethered camera at the tip (Fig. 3A). The tethered camera is positioned inside the inner membrane, and pressurized air used for eversion naturally flows within the internal channel, reducing friction between the sensor wire and the inner membrane. The point where the everting section connects with the non-everting section (Fig. 2B) allows air to passively flow, thereby minimizing friction between the camera and the internal material. A tip mount is attached to the end of the camera, preventing it from being engulfed by the membrane during retraction and maintaining a consistent focal distance from the intestinal wall for clear imaging.

This passive synchronization mechanism allows the eversion mechanism to retain its original structure, even with tethered sensors, thereby keeping the pressure required for growth low and achieving low stiffness. This enables the robot to easily conform to and efficiently grow within the intestinal environment while capturing images. Furthermore, the fixed camera-length movement method, which will be discussed later in the base section, minimizes unnecessary tension in the inner part, thereby further reducing the robot’s stiffness.

Steering

Although the colon has sharp turns, it basically forms a single continuous cylindrical path. The eversion mechanism, with its frictionless locomotion and adaptability, is well-suited to navigating these single-channel environments. Additionally, intestinal spasms, which can pose challenges during conventional flexible endoscopic procedures, may also provide helpful guidance to the SGE by offering extra adaptive support.

Given that the colon inherently guides the propulsion of the SGE, steering can primarily rely on passive, environment-aided methods. Therefore, the SGE mainly depends on adaptive steering, supported by the colon’s natural structure. This adaptive steering can be further enhanced by appropriately selecting the vine’s diameter to reduce bending stiffness. However, special scenarios may arise due to the natural folds of the colon. These folds can sometimes act as barriers, potentially causing unintended retroflection or hindering the navigation.

To address these issues, the vine should incorporate active steering methods in addition to passive steering. Therefore, the SGE is equipped with fPAMs for additional steering (Fig. 3B). The SGE features three fPAMs evenly spaced around the vine, enabling 2DoF steering, while also functioning as retraction channels, facilitating stable retraction. This steering method does not interfere with the eversion mechanism, preserving the original growth performance with low growth pressure. It can be actively actuated to enhance the robot’s curvature only when needed, complementing the passive steering that is already enabled by the robot’s low bending stiffness during growth.

Retraction: Self-retractable mechanism

In endoscopic procedures, diagnosis and therapeutic intervention usually occur during the withdrawal phase, making retraction a crucial technical function. However, due to the structural characteristics of the eversion mechanism, severe buckling can occur during retraction, leading to unintended forces being applied to the surrounding environment, which must be avoided.¹⁷

Buckling typically occurs when the inner membrane is simply pulled from the base, as its asymmetry generates a significant moment that leads to instability. To reduce this buckling moment, several approaches have been proposed, such as embedding rigid structures along the inner membrane¹⁹ or using static inflation to keep the inner membrane centered.²¹ However, these methods tend to increase the bending stiffness of the entire structure and are generally not feasible in curved anatomical paths. Fundamentally, they still rely on base-driven pulling of the inner membrane, which causes cumulative friction, especially in tortuous geometries, and limits their application primarily to straight paths. Moreover, these techniques do not account for the integration of tethered sensors within the inner channel.

Other strategies involve direct-driven mechanisms at the tip to invert the outer membrane during retraction, such as motor-driven systems¹⁷ and inchworm-like actuation mechanisms.¹⁸ These approaches can be effective in curved regions, as the tip is directly inverted rather than relying on tension in the inner membrane. However, such systems compromise the soft, morphing nature of vine robots. Embedding motors or pneumatic hoses at the distal end and proximal parts increases bending stiffness, reduces compliance, and hinders navigation through the colonic paths. Their complex hardware configurations also interfere with tethered sensor integration and are limited by inherently slow retraction speeds during clinical operation.

To overcome these limitations, we introduce a non-sealed, SR mechanism, which forms the core of the proposed SGE system (Fig. 3C). This mechanism is a direct-driven solution that does not require bulky hardware and builds upon the design proposed by Kim et al.²⁰ while addressing the specific requirements of endoscopic operation. In this design, high-velocity air is introduced through the inner membrane-side retraction channels while simultaneously being vented through outlets integrated with the steering channels. This creates a dynamic internal flow that generates tip-driven inversion by fluid momentum.

The momentum of the airflow inverts the tip directly, analogous to the behavior of a sky dancer, where fluid momentum restores the structure to an upright configuration.³⁴ Additionally, the pressure drop induced by orifice-like constriction at the tip and the 180° reversal of flow direction creates a localized low-pressure region, thereby minimizing the force required for tip inversion (Fig. 3D).^35,36 Furthermore, with the outer membrane side intentionally depressurized, the entire retraction process becomes functionally equivalent to pulling back an inflated, cylinder-like inner membrane—a configuration inherently resistant to buckling, as described by Kim et al.²⁰ This hydrodynamic approach enables buckling-free, tip-driven retraction without structural modifications to the conventional vine robot, thereby preserving both simplicity and compliance. The system is also compatible with embedded tethered sensors and capable of operating within highly curved anatomical pathways.

The proposed SGE integrates tethered sensors, steering, and retraction mechanisms directly into the everting membrane without adding hardware to the inserted section. This design minimizes system complexity while maintaining low growth pressure and stiffness. Ultimately, it enables the robot to perform essential endoscopic functions while conforming to and advancing through the intestinal tract.

Base

Eversion-based robots require a base to store the membrane material. Figure 2C shows a side view of the SGE’s base. The membrane material is stored in a straight configuration, a design consideration tailored to the limited length requirements of endoscopic applications. This design minimizes friction between the inner membrane and the sensor cable, ensuring a stable internal channel.

The SGE’s base features a pulley-like design that continuously supplies pneumatic pressure to the retraction channels while controlling the eversion length. The pulley-like design cancels out the excess length of the internal material required for the eversion, maintaining the overall length of the base equal to the vine’s extended length. Consequently, the vine’s extended length directly corresponds to the movement distance of the transporter within the base, allowing for precise length control. Furthermore, this fixed-length camera movement consistently maintains low tension on the inner membrane, as the camera length always exceeds the grown vine. This ensures that the growth force is solely applied to the membrane for eversion.

The transporter inside the base houses the camera wire, a motor for camera length control, and the structure that holds the retraction channels. The transporter’s position is controlled antagonistically by motors at both ends of the base. The camera wire maintains its total length from the transporter to the tip and compensates for length discrepancies due to curvature with a motor dedicated to camera length control. The section of the camera wire extending outside the base is coiled to accommodate adjustable lengths. Pneumatic pressure required for each function is supplied through corresponding inlets (Fig. 3). The three main functions for navigating the SGE are demonstrated in Supplementary Video S1.

Experimental Results

Resistive force during insertion

The pain experienced during the endoscopy procedure is primarily caused by the distension of the colon.³⁷ In the conventional flexible endoscopy, the colon shortening method, such as right turn, is utilized by hooking the first curve of the descending colon. To achieve this process, the FE should be pushed with sufficient force to pass over the first sharp curve of the sigmoid colon. This procedure causes a large amount of extension to the colon mesentery, and the patient’s pain significantly increases.³⁸ Furthermore, perforation most frequently occurs in the sigmoid colon during diagnostic colonoscopy.^39,40 Therefore, we compared the resistive force at the base end during insertion into the sigmoid colon between the pushing method of a conventional FE and the everting method of an SGE. The force measuring point was set at the base, as shown in Figure 4A, B, to measure the total resistive force, which is the reaction force that represents the force exerted on the colon. This indirect measurement of total resistive force was chosen to allow the free movement of the colon position, simulating the real colonoscopy situations.

FIG. 4.

Proof of concept for soft growing endoscopy (SGE) compared with conventional endoscopy. (A) Schematics of the experimental setup for (i) conventional endoscopy and (ii) SGE. (B) Photographs of the actual experimental setup corresponding to the schematic shown in (A). (C) Measured resistive forces during the rectosigmoid junction insertion process of each endoscope. (D) Normal forces applied to the intestinal walls during the rectosigmoid junction insertion process. (E) Shear forces acting on the intestinal walls during the insertion into a straightened intestinal model. (F) The average shear forces measured in (E). The shaded areas represent the 95% confidence intervals (N = 5).

The total resistive force of the pushing method increases with insertion length, as shown in Figure 4C. In contrast, the everting method exerts a lower force than the pushing method, and the force decreases as the insertion length increases. This difference arises because, in the pushing method, the FE must overcome frictional forces during insertion, with the sharp curve of the sigmoid colon further exacerbating these forces. Conversely, the everting membrane (vine) in the everting method is less affected by environmental conditions due to its frictionless, tip-growing locomotion. Please refer to the “Experimental Methods” section in the Supplementary Data for a more detailed description of the experimental procedures.

Exerted force on the colon during insertion

The exerted force on the sigmoid colon, where the endoscopic procedure primarily extends the mesentery,³⁸ was directly measured along the primary axis using a uniaxial load cell (Supplementary Fig. S1). The pushing method exerted a significantly higher force on the colon compared with the everting method (Fig. 4D), indicating that the everting method minimizes the direct mechanical load applied to the colon. This reduction in force suggests a potentially less invasive interaction with the colon wall, which contributes to improved patient comfort during the procedure.

Examining the results of the resistive force and normal force experiments together reveals a key difference: for the conventional endoscope, the large force applied near the anus was reduced at the distal end. In contrast, with the SGE, the force at the distal tip was higher than the resistive force near the anus. This can be explained by the fact that both devices experience friction at the proximal part of the intestine. However, while the conventional FE must overcome this friction to advance, the growth force of the SGE is supported by the friction, allowing it to grow forward more easily. Additionally, this growth-based propulsion mechanism does not require the force applied at the anal side to be transmitted to the distal end, giving the SGE higher resistance to endoscope looping. In other words, for the same distal force output, the SGE experiences significantly lower compressive stress on its proximal part, thus reducing the chance of looping.

Robustness on viscous colon

The viscosity of the colon is a critical factor during endoscopic procedures. A viscous colon can often result from poor bowel preparation and the evaporation of lubricant.⁴¹ A viscous colon requires excessive force to advance the endoscope, thereby increasing the patient’s discomfort.^38,42 The viscosity also prevents the colon from expanding efficiently. Although lubricants may help alleviate this issue, their effect can gradually diminish as the lubrication is removed from the shaft while it slides along the colon.^43,44 Figure 4E, F show the significant shear forces on the colon that can occur when advancing the endoscopes under these conditions. The shear force was directly measured using a uniaxial load cell (Supplementary Fig. S2). The pushing method of FE exerts drastically increased shear force without lubrication as the insertion length increases. In contrast, the everting method maintains consistently low shear force, both with and without lubrication. This demonstrates the robustness of the everting method in conditions of a viscous colon and other adhesive situations caused by poor preparation and dehydration. Although localized shear forces were found to be similarly low for the lubricated FE and the SGE, the cumulative effect of shear-induced dissipation in the FE limits distal force transmission, especially in long or curved paths. In contrast, the SGE maintains consistent tip force due to its eversion-based propagation, which avoids cumulative losses. This highlights a fundamental difference in how friction impacts system-level performance, even when local forces appear comparable.

Characterization and parameter selection of the soft growing endoscope

Figure 5 presents the characterization results of the SGE. To develop the eversion mechanism (vine) of the SGE while ensuring compliance with the intestine, we first examined the diameter of the vine. Eversion characterization was performed using a colon model to assess the success rate in passing two consecutive 90° curves (Supplementary Fig. S3). Unintended twisting and retroflection were considered failures. Instances of failure, becoming stuck in the haustra, and successful traversal through the entire colon were recorded. This experiment was based on the hypothesis that smaller diameters may lead to directional instability due to increased sensitivity to localized bending forces, which arise as reaction forces from the intestinal wall during insertion. In contrast, larger diameters may lack sufficient flexibility to conform to sharp anatomical turns. Figure 5A presents the outcome counts for five different diameters of vines. Smaller diameters were more prone to failure, while large diameters occasionally became stuck inside the colon. Diameters less than 25 mm resulted in a failure rate of over 75%; therefore, diameters greater than 25 mm were considered for further analysis.

FIG. 5.

Characterization experiments of the everting membrane. (A) Results showing the number of successful passages through a continuous 90° bend without external intervention. Membranes with diameters of 25, 28, and 31 mm exhibited a success rate of over 60%. (B) Measured growth pressure for diameters with a success rate of over 60% under simulated intestinal pressure conditions (N = 5). (C) Curvature versus restoring moment for each diameter at the measured pressures (N = 5). (D) Measured retractable length relative to the membrane diameter for three different retraction channel diameters (N = 5). (E) Characterization of the tip diameter for the optimal ratio of retraction channel and membrane diameter as determined in (D). (F) Maximum steering curvature as a function of pre-stretch ratio for the retraction (steering) channel, which maintains the diameter ratio determined in (D), before it is attached to the everting membrane (N = 3).

The internal pressure of the eversion mechanism is primarily used to generate tip force, deform the material, and overcome the friction necessary to transmit the inner material from the base to the tip. However, in a real colonoscopy scenario, additional pressure is needed to overcome the contraction forces of the colon by the spasm. To validate the relationship between the internal pressure of the vine inside the colon for growth (growth pressure) and the vine’s diameter, we conducted growth experiments inside a hydrostatically pressurized colon model (Supplementary Fig. S4). The colon model was submerged underwater to a depth of 20 cm to simulate hydrostatic pressure, replicating the spasms of the colon. The internal pressure of the vine was measured at the onset of growth under these conditions. As shown in Figure 5B, the pressure exhibits a low-slope linear trend relative to the diameter and is higher compared with the growth pressure under atmospheric conditions for generating tip force, which is primarily determined by the material characteristics of the vine.

Regarding the colon deformation that causes the patient’s pain, the bending stiffness of the endoscope, which straightens the colon’s natural curvature during inflation, is the primary factor to consider. Here, we measured the restoring moments corresponding to the given curvatures for three distinct vine diameters: 25, 28, and 31 mm (Supplementary Fig. S5). The internal pressure of the vines was set to the growth pressure values measured in the previous experiment. Figure 5C shows the restoring moment of each diameter of vines. The results show that a larger diameter corresponds to a higher restoring moment at the same curvature. Therefore, a diameter of 25 mm was selected for the vine of the SGE to maximize compliance while ensuring reliable passive steering within the colon.

The diameter of the steering channel, which also serves as the retraction channel, should be optimally selected to enable maximum steering curvature and retraction force while effectively preventing jamming within the vine during retraction. Figure 5D illustrates the occurrence of jamming at different lengths of the straight everted section for various diameter ratios. This experiment was based on the hypothesis that lower diameter ratios increase the risk of jamming due to channel contact and friction, whereas larger ratios improve retraction performance but may introduce structural instability. A diameter ratio of 1.5 between the vine and retraction channel resulted in severe jamming inside the vine, limiting retraction to only 18 cm of the everted section. In contrast, a diameter ratio of 2.0, where the retraction channels barely made contact with each other, allowed for reliable retraction up to 68 cm. At a diameter ratio of 2.5, where the retraction channels no longer made contact with each other, the SGE achieved retraction without jamming issues up to 90 cm, which was the target length for in vivo validation. Since no jamming occurred at any tested length under this condition, data markers are omitted from the graph. Since a smaller channel diameter reduces the achievable steering curvature, a diameter ratio of 2.5 was selected as the optimal ratio to prevent jamming while maintaining effective steering. Moreover, the gap between the retraction channels allowed the sensor cable to be embedded within the inner channel of the vine without the risk of jamming.

The tip mount of the SGE should occupy a minimal area at the tip to ensure maximum compliance while preventing engulfment during the retraction process. A circular shape for the tip mount is considered to accommodate the irregular tip shape caused by the wrinkles of vine during retraction. We have characterized the tip mount diameter to the selected vine diameter of 25 mm. The instances of tip engulfment during 10 cm retraction for varying tip diameters were counted (Fig. 5E). The engulfment counts drastically decreased starting from a diameter of 16 mm, which is nearly two-thirds the size of the vine, and dropped to zero at a diameter of 18 mm. However, the failure of the 16 mm case is primarily attributed to the irregular positioning of the inflated retraction channel, which can be easily resolved by inflating the vine with sufficient pressure, significantly lower than the retraction channel pressure. Consequently, a 16 mm diameter was selected for the tip mount of the SGE to ensure both compliance and reliable retraction.

Lastly, the effect of pre-stretching the steering channel prior to its attachment to the vine is examined. Figure 5F shows the maximum curvature of the SGE corresponding to each pre-stretching ratio of the steering channel. A maximum pre-stretching ratio of 1.25 (from 8 to 10 cm) was achieved, significantly enhancing the maximum curvature of the SGE up to 6.08 $\pm$ 0.09 $m^{- 1}$ , and this ratio was applied in its final design.

Phantom validation

The performance of the SGE was evaluated alongside the FE during insertion into the sigmoid colon, as deformation of the sigmoid colon is a critical factor in patient discomfort. Figure 6A, B shows the insertion of the SGE and FE, respectively. Regions outside the Sigmoid-Descending junction were faded to emphasize localized deformation, and the final image highlights the SGE’s presence inside the colon. The SGE preserved the colon’s shape more effectively, with a maximum mesentery extension of 1.11 $\pm$ 0.017 times compared with the initial state, whereas the FE exhibited substantial deformation with a 1.56 $\pm$ 0.035 times mesentery extension during the demonstration. Furthermore, while the FE required a colon shortening technique to successfully enter the descending colon, the SGE navigated into the descending colon without necessitating such a technique. Retraction, without buckling or disrupting the shape of the colon, was facilitated by the non-sealed SR mechanism and the inherent compliance of the SGE, as shown in Figure 6C. Additionally, continuous visual feedback was available during both insertion and retraction of the SGE.

FIG. 6.

Experimental validation with phantom colon model. (A) Insertion process of the soft growing endoscope (SGE) and (B) conventional flexible endoscope (FE) into the phantom colon model, respectively. The yellow circles indicate the distal tip of each endoscope, while the red boxes highlight the instants of maximum displacement during the insertion. The SGE stretched the mesentery by 1.11 times from its initial state, compared with 1.56 times for the FE (N = 5). (C) The SGE successfully retracts, conforming to the shape of the intestine without causing buckling or deformation.

In addition, the transverse colon was successfully navigated, and full retraction from the cecum to the anal passage was accomplished without buckling, while maintaining continuous camera visualization throughout the procedure (Supplementary Video S3).

In vivo validation

To validate the efficacy of the SGE in a real colon environment, we conducted in vivo testing using a canine model (Mongrel, Canis lupus familiaris). The canine model was inevitably employed to ensure a fair comparison between the SGE and FE by guaranteeing successful insertion to the cecum for both devices, particularly the FE (see Supplementary Data). The structure of the dog’s large intestine closely resembles that of humans, more so than other domestic animals.⁴⁵ Similar to humans, the dog’s colon is anatomically separated into ascending, transverse, and descending parts. Although the canine model we employed features a slightly smaller mean diameter in the large intestine compared with humans, this diameter closely approximates that of the human rectum, sigmoid colon, and descending colon, regions critical for validating the insertion of the proposed eversion mechanism.^46,47

The length of the subject’s colon was approximately 90 cm, with a diameter ranging from 3 to 4 cm. Figure 7A illustrates the experimental setup for the in vivo validation. The SGE was operated using a 6DoF interface, allowing control over growth, retraction, steering (pitch and yaw), camera position, and pressure elevation to provide additional tip force during insertion. A C-arm fluoroscopy unit was used to capture X-ray images of the SGE’s insertion and retraction processes. The experiment was repeated three times, and in all trials, the SGE successfully captured images of the ileocecal valve, which marks the end of the colon (Supplementary Fig. S6).

FIG. 7.

Experimental results for in vivo evaluation of the soft growing endoscope (SGE). (A) In vivo experimental setup. (B) Main results of the in vivo validation: (i) Intraluminal pressure measurements corresponding to varying growth pressures of the SGE, (ii) image of the ileocecal valve captured by the SGE after reaching the end of the colon, (iii) X-ray image showing the overall configuration of the conventional flexible endoscope fully inserted into the colon, and (iv) X-ray image showing the overall configuration of the SGE fully inserted into the colon. (C) X-ray overview of the SGE insertion into the colon, alongside images captured by the distal tip camera during the procedure. (D) X-ray overview and corresponding tip camera images showing the retraction of the SGE after reaching the ileocecal valve (Supplementary Video S4).

Figure 7B presents the key outcomes from the in vivo validation. Figure 7B-i depicts the internal pressure within the colon corresponding to various internal pressures applied by the SGE. The results demonstrate that, despite the variability of the SGE’s operational pressure, the pressure leakage into the colon remained relatively stable, promoting adequate luminal expansion without inducing excessive radial deformation. Figure 7B-ii shows the image of the ileocecal valve captured by the SGE’s camera during the experiment. Figure 7B-iii, iv compare colon deformation between the FE and the SGE. To navigate the curved sections of the colon, the FE inevitably required the hooking method to expand and straighten the colon into a nearly straight configuration. In contrast, the SGE conformed to the natural curvature of the colon, preserving colon’s shape during both insertion and retraction, thanks to its compliance.

Figure 7C, D illustrate the SGE’s insertion and retraction processes, along with its corresponding camera views. During the insertion to the ileocecal valve, passive steering was mainly employed, demonstrating the SGE’s adaptability for successful navigation. The entire insertion was completed in 2 min. Finally, a fast and reliable retraction was performed without altering the colon’s natural curvature or causing buckling, all while maintaining continuous camera visualization.

Discussion and Conclusion

The development of an SGE involves integrating disposable materials and innovative design principles to create a device that can navigate the gastrointestinal tract. The eversion mechanism allows the SGE to extend its length and maneuver its camera tip through the intestine without friction, eliminating the need for manual pushing and pulling of the device, which are primary sources of patient discomfort. This eversion-based, compliant, self-propelled feature is anticipated to enhance patient comfort and improve operational ease for the physician, potentially reducing the need for complex manual techniques and alleviating the physical strain that may contribute to musculoskeletal disorders in endoscopists over time.

The contributions of the SGE to the general field of vine robots lie in its comprehensive integration of all essential functions required for confined-space exploration, achieved with minimal design modifications. The proposed system enables seamless functional convergence, allowing for the independent operation of key features such as growth, a tethered sensor, retraction, and steering. This functional integration preserves the simplicity of the vine structure, enhancing its disposability while maintaining the inherent advantages of the eversion mechanism. More specifically, the everting membrane, which is directly inserted into the body, does not incorporate any rigid components or complex mechanisms. This design minimizes resistance to growth and enables growth even with low pressure, effectively reducing bending stiffness. As a result, the system maintains all the advantages of soft growing robots while satisfying the requirements for endoscopic procedures, allowing the robot to conform smoothly to the intestinal environment during both growth and retraction.

The proposed non-sealed SR mechanism is the first retraction method that enables a vine robot to stably and rapidly retract in curved environments while maintaining its compliant nature and securing a tethered sensor in its internal channel. Furthermore, the compliant mechanism of the camera tip at the distal end enables hardware-based resolution of the recognizing camera’s relative position to the everting membrane (vine) without the need for additional sensors, significantly contributing to the overall structural simplicity of the system.

Moreover, the use of soft, biocompatible materials in the construction of the vine ensures that the device can safely interact with the delicate tissues of the gastrointestinal tract.^48–50 These materials also facilitate the creation of a disposable endoscope, addressing the critical issue of infection risk associated with reusable devices. The cost-effective manufacturing process of the SGE will further make it accessible for use in various health care settings, including those with limited medical resources, and shows promise for applications in veterinary medicine as well.

Our research includes extensive testing and validation of the SGE in both phantom models and preclinical trials. The investigations aim to evaluate the device’s performance in navigating the gastrointestinal tract, providing imaging, and reducing patient discomfort during colonoscopy, where it demonstrated a significant reduction in colon deformation. Notably, this study marks the first recorded case of reaching the ileocecal valve and capturing images via a wired camera embedded in the soft growing robot in live animals. These results highlight the potential for the medical application of the eversion mechanism, demonstrating the device’s ability to meet the rigorous demands of endoscopy while significantly reducing the patient’s pain.

While this study demonstrates the feasibility of the SGE in both phantom and animal models, including surrogate pain-related data such as colon mesentery distension measured in the phantom model, clinical validation remains a critical next step. Given that pain is a key clinical outcome in colonoscopy, future work should incorporate surrogate or physiological indicators of discomfort, particularly in trials where patient-reported outcomes are available. Likewise, task load metrics and standardized procedure times were not assessed in the present study due to variations in operator technique and procedural context. Future evaluations should include these factors to more comprehensively assess usability and operator workload. Establishing appropriate quantitative and qualitative benchmarks will also be essential to evaluate the system’s clinical performance.

Beyond experimental validation, developing a theoretical framework remains an important direction for future research. A quantitative model that captures the coupled dynamics of eversion, steering, and retraction would enable more precise control and systematic performance optimization. Such a framework would also support the development of model-based controllers, which are essential for advancing toward autonomous or semiautonomous operation in clinical settings.

In conclusion, the development of the SGE represents a significant advancement in addressing the limitations of traditional FEs. By integrating the latest innovations in soft robotics, this device provides a solution that not only improves procedural efficacy but also dramatically enhances patient comfort. Its unique design offers a more adaptable and less invasive approach, which could lead to a reduction in procedure-related complications and patient recovery times. As we continue to refine and optimize this technology, its future integration into clinical practice holds great potential to revolutionize the field of gastrointestinal endoscopy. This next-generation device will enable clinicians to perform safer, more efficient procedures, making gastrointestinal health care accessible to a broader patient population. The successful adoption of this novel endoscope could lead to a paradigm shift in medical device design, setting new standards for patient-centered care. Ultimately, the SGE promises to not only improve patient outcomes but also significantly elevate the overall quality of gastrointestinal health care on a global scale.

Authors’ Contributions

N.G.K. and J.-H.R. designed the study and experiments. N.G.K. worked on methodology and designed the device and fabrication process. N.G.K., S.P., D.S., and S.L. performed prototyping, phantom model experiments, and data analysis. N.G.K. and S.P. conducted characterizations. N.G.K., S.P., D.S., S.L., H.Y., and J.K. conducted in vivo experiments and analyzed the data. H.Y. and J.K. provided medical consultation and performed procedures using commercial endoscopes. N.G.K. wrote the article with input from all authors. J.-H.R. and J.K. supervised the study, with J.-H.R. overseeing technical sections and J.K. the medical sections. All authors contributed to article editing and proofreading.

Footnotes

Acknowledgment

The authors thank the medical staff (Jean Lim, Yoon Joong Kim, and Hyemi Jee) at the Preclinical Research Center, Biomedical Research Institute, Seoul National University Bundang Hospital, for providing technical support during the in vivo experiments.

Author Disclosure Statement

The authors declare that they have no competing interests.

Funding Information

This work was supported in part by the National Research Foundation of Korea (Grant RS-2025-00554618). NAVER Digital Bio Innovation Research Fund, funded by NAVER Corporation (Grant 3720230130). SNUBH Research Fund (Grant 16-2021-0002).

Supplemental Material

References

Sinonquel

, Jans

, Bisschops

. Painless colonoscopy: Fact or fiction? Clin Endosc, 2024; 57(5):581–587.

Lee

S-H

, Park

, Lee

, et al. Colonoscopy procedural skills and training for new beginners. World J Gastroenterol, 2014; 20(45):16984–16995.

Chung

, Han

, Yoo

, et al. Patient factors predictive of pain and difficulty during sedation-free colonoscopy: A prospective study in Korea. Dig Liver Dis, 2007; 39(9):872–876.

de’Angelis

, Di Saverio

, Chiara

, et al. 2017 WSES guidelines for the management of iatrogenic colonoscopy perforation. World J. Emerg. Surg, 2018; 13:1–20.

Al-Rifaie

, Gariballa

, Ghodeif

, et al. Colonoscopy-related injury among colonoscopists: An international survey. Endosc Int Open, 2021; 9(1):E102–E109.

Mishima

, Aoki

, Hirose

. Development of pneumatically controlled expandable arm for search in the environment with tight access, In Field and Service Robotics: Recent Advances in Research and Applications. Springer: Berlin; 2006, pp. 509–518.

Blumenschein

, Coad

, Haggerty

, et al. Design, modeling, control, and application of everting vine robots. Front Robot AI, 2020; 7:548266.

Hawkes

, Blumenschein

, Greer

, et al. A soft robot that navigates its environment through growth. Sci Robot, 2017; 2(8):eaan3028.

Al Harthy

, Sadati

, Girerd

, et al. Tip-growing robots: Design, theory, application. Surgical & Interventional Engineering, 2024.

10.

Slade

, Gruebele

, Hammond

, et al. Design of a soft catheter for low-force and constrained surgery, in Proceedings of the 2017 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 24 to 28 Sept. 2017, Vancouver, Canada, pp. 174–180.

11.

, Obregon

, Heit

, et al. VINE catheter for endovascular surgery. IEEE Trans Med Robot Bionics, 2021; 3(2):384–391.

12.

Berthet-Rayne

, Sadati

, Petrou

, et al. Mammobot: A miniature steerable soft growing robot for early breast cancer detection. IEEE Robot Autom Lett, 2021; 6(3):5056–5063.

13.

Song

, Fallegger

, Trouillet

, et al. Deployment of an electrocorticography system with a soft robotic actuator. Sci Robot, 2023; 8(78):eadd1002.

14.

Dehghani

, Welch

, Pourghodrat

, et al. Design and preliminary evaluation of a self-steering, pneumatically driven colonoscopy robot. J Med Eng Technol, 2017; 41(3):223–236.

15.

Jeong

S-G

, Coad

, Blumenschein

, et al. A tip mount for transporting sensors and tools using soft growing robots, in Proceedings of the 2020 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 25 to 29 Oct. 2020, Las Vegas, NV, pp. 8781–8788.

16.

Kim

J-H

, Jang

, Lee

, et al. Origami-inspired new material feeding mechanism for soft growing robots to keep the camera stay at the tip by securing its path. IEEE Robot Autom Lett, 2021; 6(3):4592–4599.

17.

Coad

, Thomasson

, Blumenschein

, et al. Retraction of soft growing robots without buckling. IEEE Robot Autom Lett, 2020; 5(2):2115–2122.

18.

Heap

, Naclerio

, Coad

, et al. Soft retraction device and internal camera mount for everting vine robots, in Proceedings of the 2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 27 Sept. to 1 Oct. 2021, Prague, Czech Republic, pp. 4982–4988.

19.

, Szczech

, Cao

. A retractable soft growing robot with a flexible backbone, in Proceedings of the 2023 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 1 to 5 Oct. 2023, Detroit, MI, USA, pp. 2477–2484.

20.

Kim

, Seo

, Park

, et al. Self-retractable soft growing robots for reliable and fast retraction while preserving their inherent advantages. IEEE Robot Autom Lett, 2024; 9(2):1082–1089.

21.

, Mintchev

. Vine robots that evert through bending, in Proceedings of the 2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 14 to 18 Oct. 2024, Abu Dhabi, UAE, pp. 10535–10542.

22.

Greer

, Morimoto

, Okamura

, et al. Series pneumatic artificial muscles (sPAMs) and application to a soft continuum robot, in Proceedings of the 2017 IEEE International Conference on Robotics and Automation (ICRA), 29 May to 3 June. 2017, Singapore, pp. 5503–5510.

23.

Naclerio

, Hawkes

. Simple, low-hysteresis, foldable, fabric pneumatic artificial muscle. IEEE Robot Autom Lett, 2020; 5(2):3406–3413.

24.

Mendoza

, Naclerio

, Hawkes

. High-curvature, high-force, vine robot for inspection, in Proceedings of the 2024 IEEE International Conference on Robotics and Automation (ICRA), 13 to 17 May. 2024, Yokohama, Japan, pp. 3014–3021.

25.

Deglurkar

, Xiao

, Gockowski

, et al. A light-and heat-seeking vine-inspired robot with material-level responsiveness. IEEE Robot Autom Lett, 2024; 9(1):1–8.

26.

Davy

, Greenidge

, Kim

, et al. Vine robots with magnetic skin for surgical navigations. IEEE Robot Autom Lett, 2024; 9(8):6888–6895.

27.

Kim

, Greenidge

, Davy

, et al. External steering of vine robots via magnetic actuation. Soft Robot, 2025; 12(2):159–170.

28.

Shike

, Fireman

, Eliakim

, et al. Sightline ColonoSight system for a disposable, power-assisted, non-fiber-optic colonoscopy with video. Gastrointest Endosc, 2008; 68(4):701–710.

29.

Saxena

, Paul

, Haluc

, et al. Tubular locomotion and positioning using tip eversion for endoscopy. J. Med. Devices, 2020; 14(2).

30.

Pore

, Piccinelli

, de Rossi

, et al. EndoVine: Soft robotic endoscope for colonoscopy, in Proceedings of the 14th Hamlyn Symposium on Medical Robotics, London, UK, 2022, pp. 117–118.

31.

Dinkel

, Weinmann

, Pott

, et al. Pressure modulation improves locomotion of an expanding robot for colonoscopy. Curr. Dir. Biomed. Eng, 2024; 10(4):187–190.

32.

Borvorntanajanya

, Shi

, Ahmed

, et al. Model-based tracking control of a soft growing robot for colonoscopy. IEEE Trans Med Robot Bionics, 2025; 1–1:1–1.

33.

Giri

, Girerd

, Cervera-Torralba

, et al. InchIGRAB: An inchworm-inspired guided retraction and bending device for vine robots during colonoscopy. IEEE/ASME Trans Mechatron, 2025:1–12.

34.

Cros

, Romero

JAR

, Flores

. Sky dancer: a complex fluid-structure interaction, in Experimental and Theoretical Advances in Fluid Dynamics. Springer; 2012, pp. 15–24.

35.

Miller

. Flow Measurement Engineering Handbook. McGraw-Hill; 1983.

36.

Munson

, Young

, Okiishi

. Fundamentals of fluid mechanics. Oceanogr. Lit. Rev, 1995; 10:831.

37.

Elphick

, Donnelly

, Smith

, et al. Factors associated with abdominal discomfort during colonoscopy: A prospective analysis. Eur J Gastroenterol Hepatol, 2009; 21(9):1076–1082.

38.

Appleyard

, Mosse

, Mills

, et al. The measurement of forces exerted during colonoscopy. Gastrointest Endosc, 2000; 52(2):237–240.

39.

Waye

, Kahn

, Auerbach

. Complications of colonoscopy and flexible sigmoidoscopy. Gastrointest Endosc Clin N Am, 1996; 6(2):343–377.

40.

Orsoni

, Berdah

, Verrier

, et al. Colonic perforation due to colonoscopy: A retrospective study of 48 cases. Endoscopy, 1997; 29(3):160–164.

41.

Watanabe

, Hirose

, Ikegaya

, et al. Identification of lubricant viscosity to minimize the frictional impact of colonoscopy on colonic mucosa. Commun Eng, 2024; 3(1):31.

42.

Geenen

, Schmitt

Jr , Wu

, et al. Major complications of colonoscopy: Bleeding and perforation. Digest Dis Sci, 1975; 20(3):231–235.

43.

Brocchi

, Pezzilli

, Bonora

, et al. Oil-lubricated colonoscopy: Easier and less painful? Endoscopy, 2005; 37(4):340–345.

44.

Brocchi

, Pezzilli

, Tomassetti

, et al. Warm water or oil-assisted colonoscopy: Toward simpler examinations? Am J Gastroenterol, 2008; 103(3):581–587.

45.

Greaves

. chap. 8. In Histopathology of Preclinical Toxicity Studies: Interpretation and Relevance in Drug Safety Evaluation. Academic; 2011.

46.

Sadahiro

, Ohmura

, Yamada

, et al. Analysis of length and surface area of each segment of the large intestine according to age, sex and physique. Surg Radiol Anat, 1992; 14(3):251–257.

47.

Deschamps

, Humbert

, Zentek

, et al. From Chihuahua to Saint-Bernard: How did digestion and microbiota evolve with dog sizes. Int J Biol Sci, 2022; 18(13):5086–5102.

48.

De Nicola

. Permanent artificial (silicone) urethra. J. Urol, 1950; 63(1):168–172.

49.

Gunatillake

, Adhikari

. Nondegradable Synthetic Polymers for Medical Devices and Implants, in Biosynthetic Polymers for Medical Applications. Woodhead; 2016, pp. 33–62.

50.

Shakiba

, Rezvani Ghomi

, Khosravi

, et al. Nylon—a material introduction and overview for biomedical applications. Polymers Adv Techs, 2021; 32(9):3368–3383.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

29.27 MB

0.00 MB

1.95 MB