Novel shape-lockable self-propelling robot with a helical mechanism and tactile sensing for inspecting the large intestine

Figure 1 illustrates the design concept of the proposed self-propelling robot. Basically, the structure of the robot consists of a propulsion module, sensing module (figure 1(d)), and shape-locking module (figure 1(c)) integrated through a multi-sectional structure (backbone). As shown in figure 1(b), the backbone is assembled by sequencing a head link, tail link, and several waist links together along a central cord. In this structure, the length between any two neighboring links is 11 mm, and the unilateral rotation limit of each spherical joint is 30°. Moreover, an O-ring is attached to each link without a sensing module to ensure that all links are consistent in diameter and to assist the robot in obtaining sufficient friction with the inner wall of the colon.

Zoom In Zoom Out Reset image size

The propulsion module comprises the backbone and six expandable actuating balloons. The actuating balloons are distributed uniformly around the backbone at an angle to the central cord through the channels on the waist links, and the two ends of each balloon are mounted on the quick couplings of the head and tail links. The locomotion principle of the propulsion module will be discussed in section 2.2.

During colonoscopy, a stable inspection of important sections of the colon, such as lesions, is inevitably performed. Therefore, colonoscopes that can flexibly adjust their field of view and maintain a stable state are of great importance. To make the robot stable for visualization in tubular environments, a shape-locking module has been specially designed. The shape locking module includes three locking wires, three sliders, three stoppers, a locking balloon, and a locking shell. The locking wires are spirally distributed around the backbone, similar to the arrangement of the actuating balloons. One end of each locking wire is fixed to the head link, and the other end is fixed to the slider. The inflation and deflation of the locking balloon, which is glued to the stoppers, leads to the radial displacement of the stoppers. As shown in figure 1(a), there are teeth on both sliders and stoppers. Thus, inflating or deflating the locking balloon, which determines whether the stoppers are engaged with the sliders or not, can switch the robot between the locking-on and locking-off states. In the locking-off state (figure 2(a)), the sliders slide freely in the locking shell, and the locking module has little effect on the propulsion module; in the locking-on state (figure 2(b)), the stoppers, which cannot move axially, are engaged with the sliders, locking the sliders to their current positions. When the slider positions are locked, the three locking wires work together to stabilize the robot in a certain shape.

Zoom In Zoom Out Reset image size

Robots in tubular environments should avoid slippage as slippage greatly reduces the efficiency of propulsion [24]. To do so, the robot must maintain effective contact with the inner wall of the tube at all times. However, the diameter of some tubular environments changes constantly; accordingly, this changes the contact status between the robot and the inner wall of the tube. In addition, the contact force in colonoscopy should always be under a threshold value to ensure safety. In this robot, a sensing module is integrated on the periphery of the middle waist link to monitor the contact status. Thus, the robot can be adjusted in real time to avoid slippage and ensure safety. As shown in figure 1(d), the sensing module, which is mounted on the middle waist link, consists of a silicone substrate and a specially designed soft sensor based on the piezoresistive effect. The sensor corresponds to an actuating balloon; therefore, the contact status between the robot and the environment can be monitored when the actuating balloon is driven. The material between the electrodes is pressure-conductive rubber whose resistance can change with the pressure applied to it [25].

To facilitate the analysis of the propulsion module, we simplified its model to a cylindrical body with six actuating balloons and a central cord; the cord and the actuating balloons are fixed together in the body whose role is the same as the backbone. The actuating balloons are spirally distributed around the central cord, and the length between the center of each balloon and the central cord is r. We set the pitch of the spiral balloons to P₀. As shown in figure 3(a), when the length of the body is exactly one pitch, the initial length of each balloon L₀ is given by the following formula:

$$\begin{align}{L_0}^2 = {\left( {{P_0}} \right)^2} + {\left( {2\pi r} \right)^2}.\end{align} \tag{ 1 }$$

Zoom In Zoom Out Reset image size

The length of the central cord is immutable. As shown in figure 3(a), when one of the actuating balloons expands lengthwise, the shape of the body becomes helical, and the corresponding expanded balloon goes onto the outer side of the helical body. P and R represent the pitch and radius of the helical central cord, respectively. When the length of the expanded balloon is L (L ⩾ L₀), the helical deformation can be described as follows:

$$\begin{align}\left\{ {\begin{array}{*{20}{c}} {{L^2} = {{\left( {kP} \right)}^2} + {{\left[ {2k\pi \left( {R + r} \right)} \right]}^2}} \\ {{P_0}^2 = {{\left( {kP} \right)}^2} + {{\left( {2k\pi R} \right)}^2}} \\ {L = l\left( k \right)} \end{array}} \right.,\end{align} \tag{ 2 }$$

where k is a variable coefficient, which represents the number of turns of the helical body. The helical deformation can be determined when the length of the balloon is given. Hence, a third independent equation is added to provide a unique solution to the system. Generally, the specifics of l(k) are related to the physical characteristics of the propulsion module.

As shown in figure 3(b), the cylindrical model is placed in a tube of appropriate diameter in which the expansion of any balloon can cause the corresponding outer side of the helical body to come into contact with the inner wall of the tube. To illustrate the locomotive principle of the propulsion module, we took a cross-section of the helical body as the analysis object (it is consistent with the whole helical body in motion). When the six balloons (b₁ to b₆) are alternately expanded and restored in sequence, the cross-section will rotate around the central cord and roll against the tube. The cross-section rolls along a helical line on the tube inner wall rather than in a circle because the rotation axis of the cross-section is at an angle to the central axis of the tube. Hence, the whole helical body can generate a helical rotation that can be separated into a rotation around the central axis of the tube and propulsion along the axis. The velocity relationship between these motions is marked in figure 3(b).

We assume that the outer side of the helical body is in contact with the inner wall of the tube and that there is no slippage between them. If the six balloons are sequentially expanded and restored in period T, the propelling velocity can be expressed as

$$\begin{align}{V_{\text{p}}} = V\sin \theta = \frac{{\pi d}}{T}\sin \theta ,\end{align} \tag{ 3 }$$

where d is the diameter of the body and θ is the helical angle of the central cord. According to figure 3(a) and the second equation of equation (2), we obtain

$$\begin{align}\sin \theta = \frac{{2k\pi R}}{{{P_0}}} = \frac{{2\pi R}}{{\sqrt {{{\left( {2\pi R} \right)}^2} + {P^2}} }}.\end{align} \tag{ 4 }$$

Moreover, there is a relationship,

$$\begin{align}R = \left( {D - d} \right)/2,\end{align} \tag{ 5 }$$

where D is the diameter of the tube and is equal to the outer diameter of the helical body. From equations (3)–(5), the following equation is obtained:

$$\begin{align}{V_{\text{p}}} = \frac{{\pi d}}{T}\frac{{\pi \left( {D - d} \right)}}{{\sqrt {{P^2} + {\pi ^2}{{\left( {D - d} \right)}^2}} }}.\end{align} \tag{ 6 }$$

For a particular prototype whose diameter d is constant, the velocity V_p depends on the control period T, diameter of the tube D, and pitch P. However, when the diameter D is given, the pitch P of the helical body is determined according to (5) and (2), and the propelling velocity is only determined by the control period T.

A photograph of the fabricated prototype is shown in figure 4(a); it is 195 mm in length and 22 mm in diameter. A charge-coupled device (CCD) camera is mounted on the front end of the prototype. Its electric cable is also used as the central cord mentioned in the previous analysis.

Zoom In Zoom Out Reset image size

The actuating balloons (5 mm outer diameter, 2 mm inner diameter), locking balloon (6 mm outer diameter, 2 mm inner diameter), and silicone substrate were made of a biocompatible silicone rubber (Smooth-On Dragon Skin^™ 20) using a casting method. The fabrication process of the balloons is shown in appendix. The rigid parts of the prototype were fabricated using 3D printing. Nickel–titanium alloy wires with a diameter of 0.5 mm were used as locking wires.

The pressure-conductive rubber in the soft sensor is mainly composed of a silicon rubber matrix and graphite nickel-plated fillers. When the pressure applied to the conductive rubber changes, the contact condition of the conductive fillers will change, causing a change in resistance. The dimensions of the conductive rubber are 2 × 3 × 0.5 mm. It was first placed into a groove reserved on the silicone substrate. Then, two electrodes (beryllium copper) were mounted on both sides of the conductive rubber and fixed with the silicone substrate (the electric wires were pre-installed on the electrodes).

Figure 4(b) shows the configuration of the developed robotic system. In the pneumatic section, pressurized air from an air compressor is first transported to two regulators (FESTO, MS4-LR-1/4-D6-AS), which control the supplied air pressure to the propulsion module and shape-locking module. Then, several two-position, three-way solenoid valves (FESTO, MHE2-MS1H-3/2G-QS-4) are used to achieve independent control of the air supply to each balloon. It can be seen from the robot propulsion principle that the air tubes and cables will become spirally entangled during long-distance propulsion. Therefore, we added a pneumatic–electrical rotatable joint between the solenoid valves and the robot whose output ports can rotate synchronously with the robot. In the electrical section, the cables of the CCD camera and the soft sensor are first connected to the rotatable joint. Then, the soft sensor is connected to a transmitter circuit to convert its resistance into voltage that can be received by the microcontroller (Arduino UNO), and the CCD camera is directly linked to the computer through universal serial bus. To achieve efficient actuation and monitoring of the robot, we established an interface using LabVIEW^®. Through the LabVIEW program, we can set the on-off time of the solenoid valves, and the monitoring results of the contact status is displayed in real time. With this system, the robot can be effectively adjusted to actual situations.

Figure 5 shows the given control sequence of the solenoid valves for the propulsion module. In this control pattern, the robot can always maintain a helical formation during propulsion, and the six actuating balloons can be sequentially expanded and restored in period T.

Zoom In Zoom Out Reset image size

According to the analysis in section 2.2, the propelling velocity V_p of the robot is essentially determined by the control period T and the diameter of the tube D. Therefore, we tested the velocity variation with these two values. To observe locomotion in real time, the experiments were carried out in transparent acrylic tubes (available online at stacks.iop.org/SMS/30/125023/mmedia). Figure 6(a) shows the propulsion process of the robot in a tube at a constant velocity of 5 mm s⁻¹. According to [26], the inner diameter of the intestine is mainly about 25–50 mm. However, if the colon is empty and in a contracted state, which it is during most regular intestine inspections, the lumen is much smaller. In addition, the larger diameter parts are near the cecum, where regular colonoscopies do not always reach. Therefore, the tests were conducted in tubes with diameters of 26–32 mm. We measured the velocity of the robot at a series of control periods, the results of which are presented in figure 6(b). It can be seen from the figure that under the same control period, the propelling velocity increases with an increase in the tube diameter. Notably, the propulsion velocity is inversely proportional to the control period, which is consistent with equation (6).

Zoom In Zoom Out Reset image size

The relationship between the supply pressure and propelling velocity was also tested, the results of which are shown in table 1. All velocity data are the average of three tests. It can be seen from table 1 that when the supply pressure is not larger than 1.3 bar, the robot cannot propel itself because the outer diameter of its deformed helical shape is still smaller than the tube inner diameter; thus, no effective contact is achieved. For a supply pressure between 1.4 and 1.9 bar, the velocity increases with higher pressure. However, for pressures greater than 1.7 bar, the velocity increase is not obvious. For pressures larger than 2.0 bar, the balloon bursts. Because the velocity increase is not obvious and the balloons are approaching their bursting point at pressures greater than 1.7 bar, the robot was operated at a pressure of 1.6 bar.

To comprehensively evaluate the locomotion performance, the robot was tested in a flexible tube, vertical tube, and elbow tube (available online at stacks.iop.org/SMS/30/125023/mmedia). Figure 7(a) shows the propulsion process in a flexible tube with a diameter of 30 mm. The robot was able to hold up the flexible tube and advance at a velocity of 1.8 mm s⁻¹ (control period was 3 s). To simulate the environment of a human intestine, the inner wall of the flexible tube was then wetted with water. Under this condition, the propelling velocity fell to 1.4 mm s⁻¹, which is 77.8% of the velocity in the dry flexible tube. Figure 7(b) shows the propulsion process in a vertical tube whose diameter is 26 mm. When the control period was 3 s, the propelling velocity of the robot was 0.8 mm s⁻¹, which is consistent with the results in figure 6(b).

Zoom In Zoom Out Reset image size

Figure 7(c) shows the process of the robot passing through an elbow tube at a velocity of 2.4 mm s⁻¹ for which the control period is 1.5 s and the diameter of the elbow tube is 28 mm. More than 50% of the velocity was lost when the robot passed through an elbow tube compared with a straight tube. In addition, the robot can only pass through elbow tubes where the elbow radius is larger than 100 mm. This is because the combination of six balloons and three locking wires by the links reduces the flexibility of the robot body. Further steps must be taken to improve the elbow passing performance, which will be discussed in section 4.

The piezoresistive characteristics of the soft robot were first tested using the setup shown in figure 8(a), and the results are presented in figure 8(b). We can see that the piezoresistive characteristics of the soft sensor are monotonic, making it suitable for monitoring the contact force. Moreover, the curves of the four measurement cycles are the same, which indicates that the soft sensor works stably. The sensor was integrated in a ring structure (figure 1(d)) with a flat contact surface with the environment (human body). The voltage applied to the sensor is below 1 V, and the corresponding electric current through the sensor is below 10 mA. These features ensure that the sensor is physically and electrically safe for use in the human body.

Zoom In Zoom Out Reset image size

To evaluate the effectiveness of contact status monitoring, we also tested the sensing module when the robot was propelling through a tube (available online at stacks.iop.org/SMS/30/125023/mmedia). As shown in figure 8(c), when the propulsion module was not initially actuated, the contact status signal did not change significantly. At t1, the propulsion module was activated to enable the robot to navigate through the tube. At that time, the contact status signal changed with a period equal to the control period of the propulsion module. When we increased the supplied air pressure at t2, the robot could still advance effectively, but it can be seen from the contact status signal that the contact force had increased. To adjust the robot according to the contact status, we set two contact status thresholds (c1 and c2). When the signal value is lower than the threshold value c1, we deem that the robot does not have effective contact with the inner wall of the tube (inefficient); when the signal value is higher than the threshold value c2, the contact force is deemed too large (unsafe). Two indicator lights were used to visually display changes in the contact status. Through the changes in the indicator lights, we can always adjust the contact status between c1 and c2 (both efficient and safe). Currently, c2 is considered to be 1 N, which is half of the tissue manipulating force [27] and is the safe level of contact force for medical robot work in the gastrointestinal tract; c1 is 0.35 N, which was determined by our experiment trial as the force below which the robot tends to slip.

Tests of the contact force variation for different pressures and tube diameters are shown in figure 9. It can be seen there is an approximately proportional relationship between the pressure and contact force, whereas the contact force decreases linearly with tube diameter.

Zoom In Zoom Out Reset image size

The realization of the shape-locking function essentially depends on the engagement of the stoppers and sliders. To evaluate the engagement effect between them, we measured the torque when the head and tail ends of the robot rotated relative to each other using the setup in figure 10(a).

Zoom In Zoom Out Reset image size

As shown in figure 10(b), in the locking-on state, when the rotation angle is 10°, the torque can approach an approximate value of 12 Nmm (average stiffness 1.2 N mm/°); in the locking-off state, to obtain this much torque, the rotation angle needs to reach 100° (average stiffness 0.12 N mm/°). Thus, the shape-locking module gives ten times increase in rotating stiffness between the locking-on and off states.

The shape-locking module, in conjunction with the actuating balloons, can lock the shape of the robot in an arbitrary position. To demonstrate this, an evaluation experiment was performed outside the tubes (available online at stacks.iop.org/SMS/30/125023/mmedia). As shown in figure 10(c), the robot was fixed vertically on a clamp. First, two adjacent actuating balloons were inflated, making the shape of the robot helical. Then the locking balloon was inflated, and the actuating balloons were deflated. By comparison, we can see that the helical shape of the robot hardly changed. This result indicates that the shape-locking module is effective in locking the robot.

To evaluate the visual quality and integrity during colonoscopy, the robot was tested in an opaque tube with marks on its inner surface (available online at stacks.iop.org/SMS/30/125023/mmedia). Figure 11(a) shows the process of navigating the robot in the tube at a velocity of 3 mm s⁻¹, during which the camera performed a comprehensive inspection of the tube's inner surface. The three views containing the circular marks were taken from the video captured by the camera. Furthermore, as shown in figure 11(b), the robot can be adjusted and maintained in a stable position for the inspection of a target on the tube's inner surface. This task is accomplished by coordinating the shape-locking module with the actuating balloons. By adjusting the actuating balloons, a stable view of any part on the inner surface can be obtained. The results in figure 11(b) are a good validation of the shape-locking capability of the proposed robot.

Zoom In Zoom Out Reset image size