2.2. Fabrication of Soft Fabric ActuatorFigure 1 depicts the fabrication method of textile actuators. The actuators are made by combining knit fabric on the top layer and woven fabric with a strain-limiting function on the bottom layer using a Brother CS 6000i sewing machine. The sewing is performed with 301 lock stitch at four tension settings. For sewing thread, 100 percent polyester Nm 60/2 thread was used. Table 3 shows that segment types, sample numbers, and fabric combinations of 28 different actuators were created. Actuators are made up of six different segment types that are A, B, C and 3S, 5S, 7S using two different knit (K1–K2) and two different woven fabrics (W1–W2). The fabric combinations in Table 3 define which woven fabric is used in the bottom layer of the actuator and which knit fabric is used in the top layer of actuator. These actuators have asymmetry due to the difference in width-wise length (woven fabric width: 1.5 cm, knitted fabric width: 4.5 cm) and anisotropy due to the inherent elasticity difference (Figure 1A). We discovered that the joining seam strength of the actuator made of woven and knitted fabric cut asymmetrically is more durable than the joining seam strength of the actuator cut symmetrically.To make the actuator without adding segments, the woven fabric was cut into strips 15 cm long by 1.5 cm wide, and the knitted fabric was cut into strips 15 cm long by 4.5 cm wide, as shown in Figure 1A,B. The woven fabric was then reversed on the knit fabric and aligned to its right edge. On the right edge, the aligned knitted and woven fabrics were sewn with a straight stitch. After that, the same sewing process was repeated, but this time the woven fabric was aligned to the left edge of the knit fabric. After straightening the knit strip so that it was in the center of the woven strip, the upper end of the actuator was also sewn with a straight stitch, and the actuator sample was closed on three sides. Finally, the actuator’s front face was turned out. Woven and knit fabrics were cut into strips to begin the production of segmented actuators. Across the width of the knit fabric, segment stitches were made (Figure 1C). The joining process of woven fabric and segment stitched knit strip was completed in the following steps of the production of non-segmented actuators. In our study, we labeled segment types as A, B, and C, as well as 3S, 5S, and 7S. The A, B, and C segment types indicate stitch spacing between the lower and upper width-wise stitches. The joints in the human finger were taken into consideration to create the 3S, 5S, and 7S segments. The codes 3S, 5S, 7S refer to the number of stitches created separately in the upper and lower parts of the knit fabric. 2.3. Fabrication of Thermoplastic Air Bladder

The air bladder was manufactured using an impulse sealer machine (PCS 300, Brother, Nagoya, Japan), thermoplastic polyurethane film (Stretchlon 200, Fiber Glast, Brookville, OH, USA), and polyurethane pipe. The impulse sealer machine was set to the 5.5 s time setting in order to form a bond between two rectangular cut films that were stacked on top of one another. When all three sides of the rectangle cut TPU sheet were closed, the bonding procedure was complete. By inserting a polyurethane tube into the open end, air was forced into the bladder via a pneumatic system, inflating it. The actuators in our study were composed of a single pocket produced by two layers of fabric, and a tpu bladder of 18 cm in length and 2.8 cm in width was inserted into the actuator’s pocket.

2.7. Bending Angle–Pressure TestTo better understand the bending behavior of the single chamber soft actuator, the change in its bending angle in relation to the applied pressure is recorded for each actuator as it is gradually inflated. To provide consistency between actuator samples and to facilitate understanding, bending angle is widely utilized in comparison with the curvature value [16] when measuring the actuators’ bending motion.To eliminate the variation in brightness caused by the lighting conditions in the room, the actuator was placed in a closed platform that included a camera, an installation fixture, and a single-channel test rig. An air pump, a vacuum pump, an inflate valve, a deflate valve, a vacuum valve, a 4-channel relay board, an air pressure sensor, an MPR121 capacitive touch sensor controller, manual switches, and a power supply comprise the single-channel test setup [17]. Using manual switches, the test rig can inflate, deflate, or vacuum the actuator’s bladder. Due to the nature of TPUs, vacuum is the state of full suction of the air inside the bladder, which cannot be obtained when utilizing the deflate command. The test equipment can monitor and communicate bladder air pressure and actuator capacitive sensor values to the computer. During the tests, the proximal end of the actuator was secured to the installation fixture to eliminate any unwanted motion, while the distal end moved freely in a circular motion.

A digital camera (Microsoft LifeCam HD-3000, Microsoft, Beijing, China) positioned perpendicular to the actuator’s platform captured the bending action of the actuator. Three color markers were attached to the neural axis of the actuator, and their location was fixed in accordance with the finger knuckles to establish the bending angle. As a result, the extreme points are the bottom of the proximal phalanx (M1), the middle of the medial phalanx (M2), and the top of the distal phalanx (M3).

By using automated contour detection and circle-fitting algorithms operating in real time, the coordinates of each marker’s centroid in each frame of the captured video are identified. Thus, the angle formed by M1, M2, and M3, namely angle A, is calculated by the Cosine formula:

A=cos−1b2+c2−a22bc

(1)

where b is the distance between M1 and M2, c is the distance between M2 and M3, and a is the distance between M1 and M3. The bending angle is calculated as the conjugate angle to the central angle that sub-tends to the inscribed angle, A.

To demonstrate the experiment’s repeatability, the actuator’s inflation and deflation processes were repeated six times sequentially. Due to the fact that the number of frames collected per second and the pressure values obtained were varied, these values were matched using their timestamps. To produce more stable results, the initial cycle of pressure application and withdrawal was omitted. After obtaining the matched data, just the inflation component was retrieved, as only the inflation component contained valuable information on the bending motion. The average pressure value corresponding to each bending angle was calculated. This procedure was followed for all actuator types.

2.9. Radial Expansion TestIn unrestricted designs, radial expansion of the actuator walls results in an increase in its diameter and a decrease in bending angle over a wide range of pressure levels. The radial expansion of two distinct actuators is determined, and their performance is compared in relation to the applied internal pressure. To accomplish this, the actuator was put in an isolated area that was internally illuminated in order to maintain a static lightning state during the experiment. A rectangular object with a defined width and height and a different color than the actuator was positioned beneath the actuator, and the camera installed at the top of the experiment environment captured the actuator’s motion over the object. Due of the color-based method used, the experiment area was intended to contrast with the object. The outside contours of the viewable parts of the item were generated using image processing techniques. As observed in Figure 4, the object looked to be divided into three sections from the camera as a result of the actuator being placed on it. The regions covered by the right and left sections can be identified based on the object’s color. However, the region of the middle part cannot be detected because it is beneath the actuator. Given the shape of the object, the region it covers may be determined by adding the upper and lower lines of the right and left sections.The number of pixels in the region occupied by the object indicates the area of the object in the image. Using the formula in Equation (3), the pixel per square cm (PPcm2) value in the image is found using: where Aobj is area of the object in cm2 and Âobj is the number of pixels in the region occupied by the object. In other words, Âobj is the pixel area of the object in the image. The region underneath the actuator provides information about the actuator diameter. The pixel area of the region underneath the actuator is calculated as:

A^under=A^obj−A^right−A^left

(4)

where Ahidden is the pixel area of the region hidden by the actuator, Aright is the pixel area of the right part of the object, and Aleft is the pixel area of the left part of the object. Then, real area of the region underneath the actuator can be calculated using PPcm2 and Ahidden as below:

Aunder=Ahidden·PPcm2

(5)

Furthermore, the actual area of the region hidden by the actuator can be calculated as below: where d¯act is the average diameter of the actuator, and hobj is height of the object. Using both formulas, the average diameter of the actuator can be calculated as below:

d¯act=A^under·PPcm2hobj

(7)

Actuator radial expansion and the air pressure data are collected simultaneously in a real-time manner.

2.10. Capacitance MeasurementThe constant DC charge approach was utilized in this work to measure capacitive sensors. This technique charges the sensor with a constant current value over a constant time period. Thus, capacitance stores a charge, which may be calculated using Equation (8): where Q is charge amount, I is the current of the charge, and t is the duration of the charge. After the charging process, the voltage information of the sensor can be measured, and the capacitance value can be calculated using Equation (9): where C is the capacitance value of the sensor, and V is the voltage value of the sensor after the charging. Using both equations, the capacitance value is calculated as follows:

It is possible to increase the sensitivity of capacitance measurement by changing the current and charging time parameters in this technique.

Capacitance is measured using the MPR121: proximity capacitive touch sensor controller. The integrated circuit’s primary role is to detect the body capacitance using the electrodes produced for the button and keypad. The capacitance of the human body fluctuates between 50 pF and 250 pF [32], and capacitive sensors produced with textiles do not surpass this value. As a result, this integration can also be used in capacitive sensors based on textiles. Charge current and time values can be modified in software on this integrated circuit, allowing for more precise measurement of capacitances across a wide range. Additionally, the integrated circuit’s two-layer digital filter reduces noise during capacitance measurements. This integrated circuit is capable of measuring up to 12 electrodes/sensors [33]. 2.11. Control StrategyA control system is prepared for managing the operation of the prototype glove mentioned in Section 3.3. Figure 5 shows the block diagram of the developed control system. The control system consists of air flow components, control components, data components, and power components.

Components of airflow include an air pump, inflate valves, deflate valves, and pressure sensors. To initiate the actuators’ bending motion, the air pump is activated, and the inflate valve is opened. The generated air is directed to the bladders via the inflate valves, and the actuator is bent. While the actuator is bending, the deflate valve is in the closed position. It is worth noting that no two air bladders or actuators are completely identical due to manufacturing conditions. When non-identical air bladders were filled simultaneously, a homogeneous bending action in all fingers was not observed. As a result, the actuators are inflated sequentially with a 200 ms inflation duration apiece, while the entire glove forms the closing movement. If all actuators meet the stop condition, no additional air is provided to the actuator that meets the stop condition, and the air pump is shut off. The air pump and inflate valves remain closed during the relax movement of the glove, while the deflate valves are simultaneously opened. Thus, the air in the air bladder flows toward the atmosphere because its pressure exceeds that of the surrounding air. When the glove provides a stopping condition for the sensors-measured relax motion, the deflate valves are closed.

The control system consists of the following components: a relay, valve drivers, a microcontroller, a USB/TTL converter, and a computer. The microcontroller controls the actuator by receiving opening and closing signals from the computer. Additionally, it controls whether the actuator is closed or opened in response to sensor input. A USB/TTL converter connects the computer and the microcontroller. When the air pump is functioning, the microcontroller may control it through the relay, and it operates at full power. Valve drivers are used to regulate valves, as a heating problem develops when the valves are left open with full power for an extended period of time; thus, the valves are opened with maximum power for 10 milliseconds and then operated with low energy to maintain the open state.

The MPR121 touch sensor, pressure sensor, microcontroller, USB/TTL converter, and computer are the signal components. When capacitive sensors are used, the noise in the sensor values increases as the distance between the measurement circuit, and the capacitive sensor grows due to changes in the line capacitance. Thus, noise can be minimized by positioning the measuring circuit as close as feasible to the sensors [34]. To minimize the noise, the MPR121 touch sensor was placed closest to the actuators. Capacitive sensors generate feedback signals, while the actuators carry out the given commands, and the system is controlled by a closed-loop control system. According to the results in Figure 6d, the capacitive sensors show approximately linear behavior with respect to the bending angle. In this way, with the upper and lower threshold values determined for the sensor values, it can be decided whether the actuator is in the closed or open state. By scaling the upper and lower thresholds, the actuator can be closed at different angles. According to the results in Figure 6d, capacitive sensors on 5 actuators show similar behavior but operate with different sensitivity. For this reason, when the system is started, all actuators are calibrated one by one, and separate upper and lower threshold values are determined for each. These obtained upper and lower limits constitute the bending and relax stop conditions of the actuators. When an actuator reaches its upper limit in inflation, it meets the stop condition and is no longer supplied with airflow. Similarly, when the capacitive sensors reach the lower limit in the relax state, the deflate valves are closed. Pressure sensors are used to control the air pressures in the actuators and provide a backup feedback signal in case any problem occurs with the capacitive sensors. Both capacitance and pressure sensor values are collected by the microcontroller, and control signals are generated according to these sensors’ data. At the same time, the sensor data are transferred to the computer via USB/TTL converter and used to visualize the system data.

The system is powered by a 12V 16A power supply. This power supply also feeds the microcontroller, relays, and valve drivers. The pressure sensor and MPR121 are fed with appropriate voltages through the voltage regulator connected to the microcontroller. Due to the voltage difference between the microcontroller and the power supply, it is given to the microcontroller after the voltage is reduced to the appropriate level with the adjustable voltage regulator.