Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Single-Actuator-Based Lower-Limb Soft Exoskeleton for Preswing Gait Assistance

Single-Actuator-Based Lower-Limb Soft Exoskeleton for Preswing Gait Assistance Hindawi Applied Bionics and Biomechanics Volume 2020, Article ID 5927657, 12 pages https://doi.org/10.1155/2020/5927657 Research Article Single-Actuator-Based Lower-Limb Soft Exoskeleton for Preswing Gait Assistance 1 1 1 1 2 Ming-Hwa Hsieh, Yin Hsuan Huang, Chia-Lun Chao, Chien-Hao Liu , Wei-Li Hsu, and Wen-Pin Shih Department of Mechanical Engineering, National Taiwan University, Taipei 10617, Taiwan The School and Graduate Institute of Physical Therapy College of Medicine, National Taiwan University, Taipei 10617, Taiwan Correspondence should be addressed to Chien-Hao Liu; cliu82@ntu.edu.tw Received 14 December 2019; Revised 16 March 2020; Accepted 30 June 2020; Published 13 July 2020 Academic Editor: Nigel Zheng Copyright © 2020 Ming-Hwa Hsieh et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. In this research, we proposed a lower-limb soft exoskeleton for providing assistive forces to patients with muscle weakness during the preswing phase of a gait cycle. Whereas conventional soft exoskeletons employ two motors to assist each leg individually, we designed a single motor for actuation. Our design assists hip flexion for light weights and prevents some slip problems that can arise from rotary motors. The actuation mechanism was based on a pulley system that converted the power supplied by the single motor into linear reciprocating motions of a slider. When the single motor rotated, the slider moved linearly, first in one direction and then in the opposite direction. The slider pulled knee braces through cables with an assistive force of 100 N. The actuation was triggered when the system detected that the backward swing of the wearer’s thigh had ended. A prototype was designed, fabricated, and examined with 7 subjects (average age, 24). Subjects were measured while they wore our exoskeleton in power-off and power-on modes. Comparisons proved that wearing the exoskeleton caused a negligible deviation of gait, and that the soft exoskeleton could reduce metabolic cost during walking. The research results are expected to be beneficial for lightweight soft exoskeletons and integration with exosuits that provide assistive forces through the wearer’s entire gait. 1. Introduction stamina, such as walking [11] and driving [12]. Recently, lower-limb exoskeletons have attracted considerable atten- Robotic rigid exoskeletons are commonly used for various tion for providing walking assistance or load-carrying capa- bility through because they can shift the load from the applications, including action assistance, augmentation, and rehabilitation. Rehabilitation-based exoskeletons are usually wearer and pass the load to the ground [13]. However, a dis- mounted on stationary facilities such as treadmills; patients advantage of rigid exoskeletons is that they are heavy and with stroke or injuries wear these exoskeletons for gait thus require undesired metabolic expenditures [14]; addi- retraining or rehabilitation [1–5]. Various portable rigid exo- tionally, they usually impose kinematic constraints when skeletons have been developed; impaired patients wear these the wearer tries to walk [15]. Few rigid exoskeletons can to regain functional abilities. When paralyzed patients wear achieve metabolic reduction for problems such as angle assis- these exoskeletons, they can walk on their feet again [6]. Even tance [16], regardless of whether they are worn for tethered if some elderly patients lack muscle strength, they can wear or stationary activities [17]. exoskeletons to regain their grabbing and gesturing abilities Soft exoskeletons composed of artificial muscles and [7]. In addition, gear-based portable rigid exoskeletons can cable-based actuation mechanisms have become very popu- enhance the strength of healthy wearers for conducting lar in the early twenty-first century because they are light- heavy-weight tasks, such as carrying loads [8] and heavy weight and comfortable. Traditional air-powered artificial lifting [9, 10], or to conduct long-term activities that require muscles require bulky air sources and often lack crucial 2 Applied Bionics and Biomechanics One walking gait cycle actuation properties [18]; cables are widely utilized as actua- tion mechanisms for soft exoskeletons due to their high stiffness. For example, some wearable soft exoskeletons (i.e., exosuits) exploit ribbons [19] or Bowden cables to transmit tensional forces and generate assistive torques on the joints; Leg of interest thus, they can act much like real human muscles. Exosuits Heel Heel Toe Heel can provide effective assistance for hip extension [19, 20], Event strike strike off off Assisted hip flexion, and ankle plantar flexion during toe-off phase Preswing Phase Stance phase Swing phase [20–22] with large savings of energy during walking. A per- Gait 0% 10% 50% 60% 100% son walking on a treadmill at 1.5 m/s, carrying a load equal 30% of his weight, can reduce his metabolic rate by 15% if Figure 1: Illustration of the assistance provided by a lower-limb soft he wears an exosuit [23]. exoskeleton during a walking gait cycle in which assistive forces In a single gait cycle, the leg that swings backward relative were supplied for hip flexion during the preswing phase. to the body can be called the “stance leg” and the leg that swings forward can be called the “swing leg.” Most exosuits hip flexion assistance. The length, width, and height of the provide assistance during the various phases of the stance exoskeleton unit are 202, 165, and 212 mm, respectively. leg and the toe-off phase of the swing leg (Figure 1). However, The exoskeleton unit comprises a start button, an actuation the other phases of the swing leg (e.g., the preswing phase) unit, sensors, a controller, and a battery (Figure 2(c)). The are often neglected. Analyses of these phases are crucial for exoskeleton unit weighs 4.3 kg and contains a motor that patients who lack muscle strength in their lower limbs. can provide a 100 N force at either knee brace. Therefore, the aim of this study was to design a soft exoskel- Figure 3 displays the actuation mechanism of the soft eton to provide hip flexion assistance during walking for exoskeleton. The motor rotates clockwise and moves the wearers who can walk but lack either leg strength or stamina. slider to the left based on the pulley system displayed in The soft exoskeleton should be lightweight and highly com- Figure 3(a) when the actuation is triggered at the preswing pliant with human bodies to avoid imposing kinematic con- phase of the right leg. The linear motions of the slider pull straints on wearers conducting daily activities. Cables with the knee brace of the right leg by the straddle cable and con- spools are commonly used as actuators for soft exoskeletons vert the motor rotation into assistive force applied at the right because they are compact and require little space. However, knee brace. In this case, the straddle cable of the left knee because force transmission relies on the frictions between brace is slacked. For the next gait cycle, the motor must rotate the spools and cables, slip problems occur and cause drifts in a counter-clockwise manner to assist hip flexion for the left in the angular positioning of gaits. Additional space is leg by pulling the straddle cable of the left knee brace. During required for installing force sensors when cables are slacked. either preswing phase, the assistive force is applied at one In this study, a single-motor-based actuation mechanism was knee brace because the system pulls the straddle cable con- designed; this mechanism enabled the proposed lightweight nected to the knee brace, as displayed in Figure 3(b). exoskeleton and prevented the potential slip problems of rotary motors. The actuation mechanism was based on a pul- 3. Modeling and Simulations ley system that converted the power supplied by the single motor into linear, reciprocating slider motions. When the 3.1. Two-Dimensional Kinematic Model of the Human Legs. A two-dimensional (2D) kinematic model of a human leg single motor rotated, the slider moved linearly in one direc- tion to pull the appropriate knee brace with an assistive force (Figure 4) was used to analyze the mechanical power of 100 N through cables. required for hip flexion or extension during a walking gait This paper is organized as follows. Section 2 introduces cycle. The thigh, calf, and foot were modeled as linkages joint the system overview of the soft exoskeleton and its actuation at the ends, and their rotation was limited to the sagittal mechanism. Section 3 presents kinematic analyses and simu- plane. In Figure 4, o, a, b, and c represent the hip joint, knee lations to discuss the conservation of mechanical power in joint, ankle joint, and toe, respectively. The thigh, calf, and the soft exoskeleton. Sections 4 and 5 describe the designs foot lengths were l , l , and l , respectively. The central 1 2 3 and control strategy of the soft exoskeleton, respectively. masses of the thigh, calf, and foot were denoted as m , m , 1 2 Finally, Section 6 reports how the soft exoskeleton was exper- and m , respectively, located on 44% of l from o, 40% of l 3 1 2 imentally examined with several subjects to investigate the from a, and 25% of l from b, respectively. The moments of metabolic costs and the influences on the gaits. The crucial inertia at the central masses of the thigh, calf, and foot were results and conclusions are summarized in Section 7. I , I , and I , respectively. The rotation angles of the thigh, 1 2 3 calf, and foot linkages were θ , θ , and θ with respect to 1 2 3 the horizontal ground. The physical parameters of the 2D 2. System Overview model of the human leg are summarized in Table 1 [24]. Figures 2(a) and 2(b) display the proposed lower-limb soft Based on the average of the ground reaction force and gait exoskeleton comprising a sling strap, a waist belt, knee locomotion data obtained from 20 young people and 20 braces, straddle cables, and an exoskeleton unit. This device adults reported in [25], the mechanical power required for provides assistive forces to the human body during walking hip flexion and extension could be obtained by multiplying by pulling the straddle cables connected to knee braces for the hip rotation moments and the angular velocity of the 212 mm 202 mm Applied Bionics and Biomechanics 3 2 2 2 l + l − a + b Sling strap cable 1 −1 φ = cos : ð2Þ Waist belt 2l l 1 cable Waist belt hook Straddle cable Brace hook Because the direction of the assistive force was in parallel with the straddle cable, the horizontal and vertical compo- Straddle cable sheath Exoskeleton unit nents of the assistive force could be obtained based on φ as Knee brace follows: ψ = − φ − θ ð3Þ hip In the context of the assistive forces in the inverse (a) (b) dynamic analysis, the mechanical power required for hip flexion and extension while wearing the soft exoskeleton is described in the next section. 3.3. Simulation Result. To investigate the influence of the soft exoskeleton on energy conservation, the mechanical power values obtained when the soft exoskeleton was worn and not worn were obtained through the inverse dynamic method. We assumed that our device could provide a con- stant assistive force of 100 N for hip flexion assistance during (c) the preswing phase for 50% to 65% of a gait cycle. As men- Figure 2: (a) Front and (b) back views of the wearable soft tioned in the previous section, the direction of the assistive exoskeleton. (c) Exoskeleton system comprising a start button, an force varied as with the angular position of the hip (Equation actuation unit, sensors, a controller, and a battery. (1)). Figure 6 presents the simulated hip moment, angular velocity of the hip joint, and mechanical power obtained dur- walking gait cycle through the well-known inverse dynamic ing a walking gait cycle. Analytical simulations were con- method. The simulation results are described later. ducted using MATLAB (MathWorks, Massachusetts, United States). The solid and dotted lines represent the cases 3.2. Modeling of the Soft Exoskeleton. Figure 5 displays a 2D in which the soft exoskeleton is worn and not worn. In model of a human leg wearing the proposed soft exoskeleton Figure 6(c), the positive value of power indicates the required in which assistive forces are supplied at the knee braces mechanical power supplied by humans for normal gait through the straddle cables. To study the effects of wearing motions. In other words, the positive area represents the the soft exoskeleton on the mechanical power required for energy consumed by humans while walking. Figure 6(a) dis- hip flexion and extension, the assistive forces were included plays that the magnitude of hip moment required during the in the aforementioned inverse dynamic analysis. preswing phase decreased due to the assistive force provided According to the 2D geometric model, the anchor point by the soft exoskeleton. This result indicated that less hip on the waist had a relative horizontal distance of a and a ver- moment was required for swinging the leg forward. More- tical distance of b. The straddle cable with a length of l cable over, the mechanical power decreased during the preswing was connected the anchor and knee points for assisting hip phase, thus decreasing the energy required from 0.3440 to rotations. The convectional hip rotation angle θ is the hip 0.2873 J/kg per cycle. The assistive force provided by the soft angle between the thigh and the vertical axis and can be exoskeleton could provide hip flexion assistance by conserv- expressed as follows: ing 3.589 J of energy required during walking. θ = θ − 270 . Here, φ represents the angle between the hip 1 tight and the straddle cable and ψ is the angle between the 4. Designs and Assembly of Soft Exoskeleton straddle cable and horizontal axis. The orientation of the straddle cable varied with the hip rotation during walking, 4.1. Pulley System. Most cable-driven actuations of the soft thus indicating that l is a function of θ and can be skeleton tend to exhibit slip problems. Deviations in the hip cable hip rotation positions were observed due to these problems. To expressed as follows: avoid these problems, a slider with reversible linear motions rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi actuated through the pulley system was developed in this pffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 2 2 2 2 −1 l = a + b + l − 2l a + b × cos π − tan − θ : study (Figure 7). All rotating components were installed on cable 1 1 hip an 8 mm thick acrylic board base (Figure 7(a)). A brushless ð1Þ DC motor and two timing belt pulleys were mounted on the rear side of the base, and four pulleys, including guiding With variation in the orientation of the straddle cable, the and driving pulleys, were mounted on the front side of the angle φ varied as the hip rotated and can be expressed as base. The two timing belt pulleys formed a gear train with follows: the gear ratio of 3 to drive the pulley for scrolling a nylon 165 mm 4 Applied Bionics and Biomechanics AF Pulley Motor Slider Load cell AF Knee brace (a) (b) Figure 3: Illustration of the actuation mechanism. (a) When the motor rotates, the slider moves through the pulley system and provides assistive forces for hip flexion by pulling the straddle cables of the knee braces and slacking the straddle cable of the other knee brace. The motor rotates reversely for providing hip flexion assistance to the other leg. (b) During the preswing phase, the assistive force is applied at one knee brace because the motor pulls a straddle cable. Sagittal plane Hip joint 𝜃 m , I 1 1 Extension Flexion Thigh Knee joint m , I 2 2 l Calf m , I 3 3 Ankle joint Foot (a) (b) Figure 4: Two-dimensional kinematic model of a human leg. thread whose both ends were attached to the two sides of the served a crucial role in precisely positioning the slider. slider (Figure 7(b)). When the motor rotated, the timing belt Because the nylon thread exhibited a large tensional stiffness pulley pulled the nylon thread and actuated the slider to value, small assembly misalignments or structure deforma- move in one direction through the driving and guiding pul- tions could easily slack the nylon thread. Although the thread leys. Because the pulley system was operated through the had been wrapped multiple times, slips occurred when the nylon thread and the end of the nylon thread was not fixed nylon thread slacked. Therefore, a pretension mechanism on the driving pulley, friction forces between the nylon was used to maintain the tightness of the nylon threads and thread and the driving pulley were key factors that caused slip was mounted on the slider (Figure 7(b)). To simplify the problems at the nylon thread. To prevent these slip problems, assembly, a torsional spring was incorporated in the slider the driving pulley was wrapped with eight turns of nylon with two ends of the spring tangled with each of its arms thread. The number of turns was limited because a high (Figure 8). The spring was pinned on the slider at its center. number of turns of the threads could increase the frictional The spring could rotate with a small angle such that a forces of the rotating pulleys. Without self-locking effects 1 mm position deviation of the slider was allowed. The spring and a relative low gear ratio of the timing belt pulley, revers- was twisted when the slider was loaded, thus allowing ible motions of the slider were achieved. An advantage of the another 1 mm position deviation. Two screws and nuts were linear slider was that the positioning drift could be eliminated used for each end of the nylon thread to fix both ends firmly, by tracking the positions of the linear sliders by photointer- which are marked as 1a, 1b, 2a, and 2b in Figure 8. Before the rupters, as described later. nylon thread was tightened, the nylon thread was wrapped with one turn on each screw in a clockwise manner on 1a 4.2. Pretension Mechanism. In addition to the friction coeffi- and 2a and in a counter-clockwise manner on 1b and 2b. Then, the nuts on 1a and 2a were screwed to hold and tighten cients and wrapping turns, the tensions in the nylon threads Applied Bionics and Biomechanics 5 Table 1: Physical parameters of the human legs [23]. Thigh Calf Foot l l l 41 cm 41 cm 20 cm 1 2 3 m m m 7.7 kg 3.1 kg 0.8 kg 1 2 3 2 2 2 I I I 1093 kg/cm 406 kg/cm 31 kg/cm 1 2 3 25% of l CM to “o” 44% of l CM to “a” 40% of l CM to “b” 1 2 3 AF during walking. The rest of the strap was wound on the 3D- Anchor point on waist printed connectors. Hip joint Straddle 5. Sensors and Control Strategy cable 5.1. Load Cells. Instead of using commercial load cells, a com- hip pact twin pull-type load cell was designed for the soft exo- cable l skeleton, as illustrated in Figure 10. In Figure 10(a), the load cell connected the heads of the straddle cables and the slider for pulling the knee braces; because the load cell had been designed to be compact, it only occupied a small vol- ume. Finite element simulations were conducted for the x Knee joint designs of the load cell in Comsol commercial software. The simplified quartered model based on two symmetric sur- Figure 5: Two-dimensional modeling of a human leg wearing the faces is presented in Figure 10(b). Because the maximum soft exoskeleton in which the assistive forces were applied at the assistive force was 100 N for the soft exoskeleton, the load cell knee joint by pulling the straddle cable. AF: assistive forces. was designed to measure the maximum assistive force of 200 N with a safety factor of 2. When a straddle cable applied a tensile assistive force of 200 N in the x-direction, the force the nylon thread. Finally, the nuts were screwed on 1b and 2b was applied at the contact point with the x-component of to fix the nylon thread. As the nylon thread undergoes ten- 100 N and y-component of 65.5 N due to the contact angle sion, the friction force between the nuts and the nylon thread of 33.23 between the load cell and head of the straddle cable. tightens the nut further due to the screw and nut mechanism. We assumed the force was exerted on the fillet surface with a Thus, the clamp of the nylon thread does not loosen, even radius of 0.5 mm, which was the surface between the head of after prolonged stress. the straddle cables and load cells. The load cell was made of stainless steel; it was designed to endure a maximum yield 4.3. Anchors and Connectors. We used straddle cables to stress no greater than 200 MPa. The simulation results are transmit assistive forces from our device to the knee braces shown in Figures 10(c) and 10(d). and then to the legs of the exosuit wearer. The connections The maximum stress was 125 MPa and was 37.5% between the cable and the knee braces were designed to be smaller than maximum yield stress, thus indicating that the easily adjustable for different wearers. Because the straddle load cell could withstand a force of 20 kgw. The maximum cables lacked flexibility and adjustability and could not be strain occurred on the upper surface of the load cell at a dis- tied with a knot on the knee braces, Velcro was used to tance of a 14 mm from the center of the load cell in the lon- adhere the straddle cables to the knee braces for ensuring gitudinal direction. This phenomenon caused a small reliable attachments. elongation of 2.04 μm, and a fine resolution could not be Figure 9(a) shows the anchor of the soft exoskeleton attained for the general metallic foil-type strain gages. There- composed of the slot for holding the cable sheath and the fore, strain gages with a large gage factor (GF) as that of open slit for attachment to the waist belt. Since the anchor semiconductor-type strain gages should be used for our attached to the waist close to the hip, walking might cause device. Multiple strain gages (Kyowa: KSN-2-120-E3-16) discomfort to the pelvic region; to maximize comfort, the with a GF of −105 were implemented using two quarter- 3D-printed anchor had a support wing filled with foam that bridge Wheatstone bridges for the twin load cell. The load distributed pressure on the hip. Figure 9(a) shows the 3D- cell was fabricated and calibrated experimentally at a mea- printed connector composed of an M4 screw, an M4 nut, surement range of 0 N–100 N and a resolution of 1.76 N and spacers. By threading the straddle cables through the (Figure 10(e)). Although the load cell exhibited a fine resolu- spacers and clamping it with the screw and the nut, we fixed tion and good linearity in the force measurements, the drift the 3D-printed connector tightly with the straddle cables. effect due to the strain gages and the increase in the temper- Then, the Velcro strap on the knee brace was wound with a ature of the circuit were not negligible. Figures 10(f) and buckle to ensure that the Velcro strap would remain fastened 10(g) display the drift effects of the left and right sides of 6 Applied Bionics and Biomechanics (a) 0.5 −0.5 0 20 40 60 80 100 7.5 5 (b) 2.5 −2.5 −5 0 20 40 60 80 100 Insufficient energy (c) Surplus energy −2 0 20 40 60 80 100 Gait cycle (%) 100 N constant force is applied Figure 6: Simulated (a) hip moment, (b) angular velocity of the hip joint, and (c) mechanical power during a walking gait cycle. Here, the solid and dotted lines represent the cases in which the soft exoskeleton is not worn and worn, respectively. A constant assistive force was applied during the preswing phase for conserving power. X = 0 Photo interrupter s Pretension mechanism Linear slider Stroke Guiding pulleys Driving pulley Timing belt pulley Motor ⁎⁎ Tensioning pulley (a) (b) Figure 7: Illustration of the power transmission based on the slider and the pulley system in the (a) top and (b) side views. the load cell, in which the measured force varied and satu- phase. Moreover, a linear relation existed between the length rated as time increased. Therefore, the measured data of load of cable l and the hip rotation angle θ , which is insen- cable hip cells were used as trigger signals in our preliminary work sitive to the changes in the geometric parameters of l , a, because accurate values of the assistive forces could not be and b, (Figure 4). Equation (1) simplified the linearization provided. method with a small variation in the hip rotation angle: 5.2. Coordinate Mapping. Because the actuation mechanism l ≈ 0:1732θ + c ð4Þ cable hip 1, converted the motor rotations to the linear motions of the slider, the coordinates of the hip angular positions were mapped to linear positions set on the sliders. The hip rotation where c is a constant based on the thigh length. By defining ° ° angles were in the range of −10 to 10 during the preswing c = l at θ =0, the mapped coordinate of the slider X 1 cable hip s Power (W/kg) (rpm) M hip (Nm/kg) Applied Bionics and Biomechanics 7 Clamp for nylon thread photointerrupters were mounted along the stroke with the mapped positions of X = −4:5, −2,−1, 0, 1, 2, and 4:5cm to track the positions of the linear motions of the slider. 1a Figure 12 displays the flowchart of the control strategy that 1b 2b 2a was realized for different stages—idle, wearer-driven pulling, actuation, and brake stages. They were determined using the motor rotational speed measured from the embedded speed sensor of the motor. If the motor rotation was higher than Nylon thread Torsional spring Turning point 800 rpm, then brakes were applied automatically and deceler- ated until the rotation speed was lower than 50 rpm. The Figure 8: Diagram of the pretension mechanism and fixation of motor rotational speed was zero while walking. Moreover, nylon threads mounted on the slider. the soft exoskeleton was in the idle stage, and the slider was at the position of X = −1or1cm at which the straddle cables of both legs were slacked. The load cells were enabled to (a) monitor the tensile forces of the straddle cables connected to the knee braces. During the end period of the backward swinging of legs, the slider was moved away from the posi- tions of X =±1 cm and the soft exoskeleton entered the wearer-driven pulling mode. At the beginning of the pre- swing phase, one side of the load cell detected a high decrease in the measured force when the leg swung forward. The actu- ation mechanism was triggered when the motor begin to rotate and thus the slider began to move linearly to provide hip flexion assistance; this period can be known as the actu- (b) ation period. When the slider passed the position of X =0 cm, the soft exoskeleton entered the brake stage and the motor rotated in the opposite direction to decelerate the lin- ear motion of the slider until the slider stopped at the posi- tion of X =±1cm. Then, the slider remained at the position of X =±1cm and the soft exoskeleton entered the idle stage until the next actuations. The photointerrupters at X = −4:5, −2, 2, and 4:5cm were designed to ensure that the slider returns to the position of X =0 cm in our original Figure 9: (a) Anchors and (b) connectors mounted on the waist. design, which is not conducted in the present study. 5.4. Actuation Trajectory. The motion trajectory of the linear was defined as the origin at which X =0 cm when the slider slider while wearing the soft exoskeleton is shown in was at the midpoint of the stroke at length of 90 mm Figure 11 based on the control strategy. At the beginning of the preswing phase, that is, after the heel of the right leg (Figure 7(b)). Based on the above condition, the formula of X can be presented as follows: strikes the ground (t =0 s), the actuation mechanism was activated due to a rapid decrease in the force measured force X =0:1732θ ð5Þ using the load cell and an assistive force was provided at the s hip: right knee by pulling the straddle cable through the linear movement of the slider by the counter-clockwise rotation of The thigh motions were projected to the mapped coordi- the motor. When the slider passed the position of X =0 cm nates by substituting θ into Equation (5). Here, the tar- s hip ° , the soft exoskeleton entered the brake stage and the motor geted hip extension or flexion angles in the range of −10 to rotated in a clockwise manner to decelerate the slider until 10 were mapped to the X range of −2 to 2 cm for the the slider stopped at the position of X = −1cm. Subse- mapped coordinates. We assumed that the walking gait of a quently, the soft exoskeleton entered the idle stage and the normal individual was bilaterally symmetrical in terms of slider remained at the same position of X = −1cm at which the phase and motion. Therefore, the slider motion projec- the straddle cables of both legs were slacked. Similarly, at tion of the left thigh X was a mirrored version of the right the end period of the backward swinging of the left leg, the thigh X with a 50% phase lag during gait cycle. The slider slider was pulled passively and moved away from the position motion projections for both thighs when the soft exoskeleton of X = −1cm. Then, at the beginning of the preswing phase was not worn are displayed in Figure 11. In this figure, the of the left leg, the actuation mechanism was triggered and the dotted line represents the right leg and the dotted–dashed motor rotated in a clockwise manner to provide an assistive line represents the left leg. force at the left knee for hip flexion assistance. At this point, 5.3. Control Strategy. A position-based control strategy was the slider moved to the position of X =1cm due to the used for the linear actuation mechanism in this study. Seven motor and pulled the straddle cable connected to the left R3.65 8 Applied Bionics and Biomechanics 1.25×10 (b) −2 ×10 ×10 1.5 35 0.5 1.2 𝛷 6 0.8 −3 ×10 0.6 0.4 0.6 0.8 0.2 12 1.2 −2 ×10 5.77×10 Position of maximum strain −4 6.31×10 −4 on upper surface ×10 −2 6 (a) 1.5 ×10 0.5 (d) −3 ×10 3 0.6 Fillet 0.8 1.2 Cross section for force applying −2 ×10 (fix boundary) z −7 3.95×10 (c) (e) (f ) (g) Figure 10: FEM simulations of the load cell. (a) Geometrical dimension (unit: mm). (b) Simplified quartered model. (c) Principal stress distributions. (d) Principal strain distributions. (e) Photograph of the load cell. (f, g) Drift effects of the left and right sides of the load cell. AF AF knee. When the slider passed the position of X =0 cm, the soft exoskeleton entered the brake stage and the motor rotated in a counter-clockwise manner to decelerate the Driven passive slider until the slider reached the position of X =1cm. Brake Heel strike Then, the slider remained at the position of X =1 cm Right leg and the soft exoskeleton entered the idle stage until next gait cycles. Idle −2 6. Experimental Results and Discussions Actuation −4 6.1. Metabolic Energy Conservation. To evaluate the capabil- Left leg −6 ities of conserving energy when a wearer of the soft exoskel- 0 0.2 0.4 0.6 0.8 1 eton walks, metabolic tests were conducted using a treadmill Time (s) machine (Figure 13(a)). Seven participants with ages ranging from 23 to 24 years, height ranging from 171 to 180 cm s (average = 174 cm, standard deviation ðSDÞ =3:16 cm), and weight ranging from 67 to 72 kg (average = 70 kg, SD = 1:91 Figure 11: Illustrations of the motion projections of the mapped kg) participated in the tests. The treadmill was setup at a con- coordinate for the cases in which the soft exoskeleton is worn and stant walking speed of 4.3 km/h. The metabolic rates of the not worn while walking. The assistive forces were provided during subjects were measured and recorded using a portable the preswing phases for each leg. pulmonary gas exchange measurement system (K4b2, COSMED s.r.l., Rome, Italy) during the test. The tests included four scenarios: (a) wearing the knee braces only X (cm) 6 12 Applied Bionics and Biomechanics 9 Idle Reading Start Open load cell Big surge ? load cell value Photo T F interrupter X = ±1 input Motor running Speed input output Speed>800 Photo rpm interrupter X = 0 input s T Actuation Brake Brake start Close Motor running Speed< load cell stopping 50 rpm Figure 12: Flowchart of the control algorithm. Pulmonary gas exchange measurement system Exoskeleton −10 −20 Treadmill −30 Loaded Power off Power on (a) (b) Figure 13: (a) Photograph of the metabolic tests conducted while walking on a treadmill. (b) The measured statistical distributions of the metabolic rate reduction for the cases in which loads are carried, power is switched off, and power is switched on. The error bar indicated one standard deviation. The metabolic rates were normalized to the case in which only the knee braces were worn (i.e., without wearing the soft exoskeleton). (i.e., free walking without wearing the soft exoskeleton), (b) worn when it is switched off, and worn when it is switched wearing the soft exoskeleton when the straddle cables were on. The metabolic rate reduction is expressed as follows: loosened (i.e., carrying loads), (c) wearing the soft exoskele- ton when it is switched off, and (d) wearing the soft exoskel- MR − MR m f eton when it is switched on. For the scenarios (c) and (d), a : ð6Þ MR 2 kgw pretensioned force was applied while wearing the soft exoskeleton. Each test lasted for 7 min. Because our device was a prototype that had not been optimized for weight Here, MR is the metabolic rate when the soft exoskele- and comfort, we focused on the effects of the assistance and ton is worn for three cases and MR is the metabolic rate ignored the influence of carrying loads. To ensure that the obtained without wearing the soft exoskeleton. The negative soft exoskeleton could be triggered appropriately and could values represented that energy was conserved due to the actuate correctly, the straddle cables were pretensioned with assistance provided by the exoskeleton, and the positive a 2 kgw force while a subject was wearing the soft exoskele- values presented that energy consumption occurred when ton. Moreover, the anchors on the waist were adjusted such loads were carried. These results demonstrated that the that the cables were in parallel with the sagittal plane. metabolic rate could be reduced further due to the assistive Figure 13(b) displays the measured statistical distribu- forces provided by our device while walking, thus indicating tions of the metabolic rate reduction for the cases in which that the proposed soft exoskeleton can conserve energy when the soft exoskeleton was worn with loosened straddle cables, the wearer walks. The deviations were due to the Metabolic rate reduction (%) 10 Applied Bionics and Biomechanics (a) Accelerated −10 −20 0 20 40 60 80 100 Phase delay Gait cycle (%) Power on Power off Free (b) Power on Power off Free 30 30 30 20 20 20 10 10 10 0 0 0 𝜃 𝜃 𝜃 −10 −10 −10 −20 −20 −20 −30 −30 −30 0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100 Gait cycle (%) Gait cycle (%) Gait cycle (%) (c) (d) (e) Figure 14: (a) Photograph of the gait motion measured using the video camera. (b) Averaged gait motions for the cases in which the soft exoskeleton is worn with its power switched on, is worn with its power switched off, and is not worn. An assistive force was provided by the soft exoskeleton during the preswing phase. (c, d, e)All gait motions for each scenario. track the positions of three markers. The angle formed by unfamiliarity in using the exoskeleton or lack of soft exoskel- eton training. the markers were measured, transformed, detrended, and finally separated through every two heel strikes in one gait 6.2. Impacts on Gaits. To investigate the effects on gait cycle. motions when the soft exoskeleton is worn, gait measure- Figure 14 displays the measured gait motions of the sub- jects. The solid lines represent that the soft exoskeleton was ments were conducted for the same subjects. Subjects wore a soft exoskeleton and walked along a straight line with three worn and was powered on, and the dashed lines represent markers attached at their shoulders, hips, and knees. More- that the soft exoskeleton was worn and was powered off. over, the angular rotation positions of the hip and knees were The dotted lines represent that only the knee braces were recorded using a video camera located at a distance of 5.3 m worn (i.e., free walking). The three curves presented in in front of the straight line (Figure 14(a)). The experiments Figure 14(b) were obtained through fitting the curves of all were conducted for the participants under three scenarios—- the gait motions for the scenarios presented in wearing the soft exoskeleton when it is switched on, wearing Figures 14(c)–14(e). A phase lag from 35% to 56% is dis- the soft exoskeleton when it is switched off, and wearing knee played in Figure 14(b); subjects attain the maximum hip braces only (i.e., free walking without wearing the soft exo- extension while wearing the device when it is switched off rel- skeleton). In the first two scenarios, a 2 kgw pretensioned ative to the free walking case. This behavior was observed force was applied while when the soft exoskeleton was worn. because that the slider remained at the last position after pull- The gait patterns were recorded at a frame rate of 30 fps. Each ing and could not return to the central point when the device frame was analyzed with image processing techniques to was switched off. When the soft exoskeleton was switched on, (deg.) hip (deg.) hip (deg.) hip (deg.) hip Applied Bionics and Biomechanics 11 the damping effect was minimized because the slider could References move and return to the central position after being pulled [1] C. H. Wu, H. F. Mao, J. S. Hu, T. Y. Wang, Y. J. Tsai, and W. L. by the subject. Therefore, the phase lag of the maximum Hsu, “The effects of gait training using powered lower limb hip extension was eliminated when the soft exoskeleton was exoskeleton robot on individuals with complete spinal cord switched. Moreover, hip rotation accelerated in the preswing injury,” Journal of NeuroEngineering and Rehabilitation, phase from 35% to 56% due to the hip flexion assistance pro- vol. 15, no. 1, p. 14, 2018. vided by the soft exoskeleton (Figure 14(b)). Because no [2] S. Jezernik, G. Colombo, T. Keller, H. Frueh, and M. Morari, obvious changes were observed in the gait, the proposed soft “Robotic orthosis Lokomat: a rehabilitation and Research exoskeleton is concluded to have at most a slight influence on Tool,” Neuromodulation, vol. 6, no. 2, pp. 108–115, 2003. the gait motions. [3] K. S. Banala, S. K. Agrawal, and J. P. Scholz, “Active leg exo- skeleton (ALEX) for gait rehabilitation of motor-impaired patients,” in 2007 IEEE 10th International Conference on Reha- 7. Conclusions and Future Work bilitation Robotics, Noordwijk, Netherlands, June 2007. A compact lower-limb soft exoskeleton was presented in this [4] J. Bae, S. M. M. D. Rossi, K. O'Donnell et al., “A soft exosuit for study for providing hip flexion assistance to walking wearers patients with stroke: feasibility study with a mobile off-board with low muscle strength. A lightweight exoskeleton was actuation unit,” in 2015 IEEE International Conference on Rehabilitation Robotics (ICORR),, Singapore, Singapore, designed to minimize slip problems; the soft exoskeleton August 2015. was designed to actuate both legs through a single actuator by converting the motor rotations into linear reciprocating [5] J. L. Contreras-Vidal, N. A. Bhagat, J. Brantley et al., “Powered exoskeletons for bipedal locomotion after spinal cord injury,” motions of the slider through a pulley system. According to Journal of Neural Engineering, vol. 13, no. 3, p. 031001, 2016. the simulation results, the exoskeleton was capable of con- [6] A. Esquenazi, M. Talaty, A. Packel, and M. Saulino, “The serving the energy required during the preswing phase of a rewalk powered exoskeleton to restore ambulatory function gait cycle. A prototype was fabricated, assembled, and exper- to individuals with thoracic-level motor-complete spinal cord imentally examined through metabolic rate tests. The results injury,” American Journal of Physical Medicine & Rehabilita- revealed that our device could reduce the metabolic cost of tion, vol. 91, no. 11, pp. 911–921, 2012. walking and exerted at most a slight influence on gait [7] P. Polygerinos, K. C. Galloway, S. Sanan, H. Maxwell, and C. J. motions, thus indicating that the proposed soft exoskeleton Walsh, “EMG controlled soft robotic glove for assistance dur- could conserve energy during hip flexions. ing activities of daily living,” in 2015 IEEE International Con- Some future studies are under investigation. Because the ference on Rehabilitation Robotics (ICORR), Singapore, actuation speed of straddle cables is limited, the current actu- Singapore, August 2015. ation mechanism cannot be used to attain a high walking [8] C. J. Walsh, K. Endo, and H. Herr, “A quasi-passive leg exo- speed. Moreover, the load cell has a significant drift effect skeleton for load-carrying augmentation,” International Jour- because of the temperature rise in the strain gage and nal of Humanoid Robotics, vol. 4, no. 3, pp. 487–506, 2011. circuitry. This problem can be overcome by finely soldered [9] W. Micheal and W. Michael, “Lower extremity exoskeleton circuitry and temperature compensations. Moreover, the reduces back forces in lifting,” in ASME 2009 Dynamic Systems measured load information can be utilized for feedback con- and Control Conference, Hollywood, January 2009. trol. Because the soft exoskeleton was designed to provide [10] K. Yamamoto, M. Ishii, K. Hyodo, T. Toshimitsu, and assistance for the preswing phase of a gait cycle, our device T. Matsuo, “Development of power assisting suit,” JSME Inter- could be integrated with other exosuits to provide assistance national Journal Series C Mechanical Systems, Machine Ele- for entire gait cycles in the future. The proposed single ments and Manufacturing, vol. 46, no. 3, pp. 923–930, 2003. actuator-based actuation mechanism is expected to be [11] Y. Hiki, Z. Sugawara, and J. Ashihara, “Walking assist device,” beneficial for realizing a compact and lightweight soft p. B2, 2013, U. S. Patent 8603016. exoskeleton. [12] B. L. Shields, J. A. Main, S. W. Peterson, and A. M. Strauss, “An anthropomorphic hand exoskeleton to prevent astronaut hand fatigue during extravehicular activities,” IEEE Transactions on Data Availability Systems, Man, and Cybernetics - Part A: Systems and Humans, vol. 27, no. 5, pp. 668–673, 1997. The measured data used to support the findings of this study are available from the corresponding author upon request. [13] A. B. Zoss, H. Kazerooni, and A. Chu, “Biomechanical design of the Berkeley lower extremity exoskeleton (BLEEX),” IEEE/ASME Transactions on Mechatronics, vol. 11, no. 2, Conflicts of Interest pp. 128–138, 2006. [14] G. J. Bastien, P. A. Willems, B. Schepens, and N. C. Heglund, The authors declare no conflicts of interest. “Effect of load and speed on the energetic cost of human walk- ing,” European Journal of Applied Physiology, vol. 94, no. 1-2, pp. 76–83, 2005. Acknowledgments [15] A. Schiele and F. C. T. van der Helm, “Influence of attachment This work was supported by the Ministry of Science and pressure and kinematic configuration on pHRI with wearable Technology, Taiwan, under grant numbers 106-2221-E- robots,” Applied Bionics and Biomechanics, vol. 6, no. 2, 002-133-MY2 and 108-2221-E-002-165. pp. 157–173, 2009. 12 Applied Bionics and Biomechanics [16] L. M. Mooney, E. J. Rouse, and H. M. Herr, “Autonomous exo- skeleton reduces metabolic cost of human walking during load carriage,” Journal of NeuroEngineering and Rehabilitation, vol. 11, no. 1, p. 80, 2014. [17] Y. Ding, I. Galiana, A. Asbeck, B. Quinlivan, S. M. M. De Rossi, and C. J. Walsh, “Multi-joint actuation platform for lower extremity soft exosuits,” 2014 IEEE International Conference on Robotics and Automation (ICRA), 2014, Hong Kong, China, June 2014, 2014. [18] Y.-L. Park, B.-r. Chen, D. Young et al., “Bio-inspired active soft orthotic device for ankle foot pathologies,” in 2011 IEEE/RSJ International Conference on Intelligent Robots and Systems, San Francisco, CA, USA, September 2011. [19] A. T. Asbeck, K. Schmidt, and C. J. Walsh, “Soft exosuit for hip assistance,” Robotics and Autonomous Systems, vol. 73, pp. 102–110, 2015. [20] A. T. Asbeck, K. Schmidt, I. Galiana, D. Wagner, and C. J. Walsh, “Multi-joint soft exosuit for gait assistance,” in 2015 IEEE International Conference on Robotics and Automation (ICRA),, Seattle, WA, USA, May 2015. [21] A. T. Asbeck, R. J. Dyer, A. F. Larusson, and C. J. Walsh, “Bio- logically-inspired soft exosuit,” in 2013 IEEE 13th Interna- tional Conference on Rehabilitation Robotics (ICORR),, Seattle, WA, USA, June 2013. [22] A. T. Asbeck, S. M. M. De Rossi, K. G. Holt, and C. J. Walsh, “A biologically inspired soft exosuit for walking assistance,” The International Journal of Robotics Research, vol. 34, no. 6, pp. 744–762, 2015. [23] F. A. Panizzolo, I. Galiana, A. T. Asbeck et al., “A biologically- inspired multi-joint soft exosuit that can reduce the energy cost of loaded walking,” Journal of NeuroEngineering and Rehabilitation, vol. 13, no. 1, p. 43, 2016. [24] C. E. Clause, Weight, volume, and center of mass of segments of the human body, Antioch College Yellow Springs, Ohio, 1969. [25] G. Bovi, M. Rabuffetti, P. Mazzoleni, and M. Ferrarin, “A multiple-task gait analysis approach: kinematic, kinetic and EMG reference data for healthy young and adult subjects,” Gait & Posture, vol. 33, no. 1, pp. 6–13, 2011. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Applied Bionics and Biomechanics Hindawi Publishing Corporation

Single-Actuator-Based Lower-Limb Soft Exoskeleton for Preswing Gait Assistance

Loading next page...
 
/lp/hindawi-publishing-corporation/single-actuator-based-lower-limb-soft-exoskeleton-for-preswing-gait-zqU1MSqWGJ
Publisher
Hindawi Publishing Corporation
Copyright
Copyright © 2020 Ming-Hwa Hsieh et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
ISSN
1176-2322
eISSN
1754-2103
DOI
10.1155/2020/5927657
Publisher site
See Article on Publisher Site

Abstract

Hindawi Applied Bionics and Biomechanics Volume 2020, Article ID 5927657, 12 pages https://doi.org/10.1155/2020/5927657 Research Article Single-Actuator-Based Lower-Limb Soft Exoskeleton for Preswing Gait Assistance 1 1 1 1 2 Ming-Hwa Hsieh, Yin Hsuan Huang, Chia-Lun Chao, Chien-Hao Liu , Wei-Li Hsu, and Wen-Pin Shih Department of Mechanical Engineering, National Taiwan University, Taipei 10617, Taiwan The School and Graduate Institute of Physical Therapy College of Medicine, National Taiwan University, Taipei 10617, Taiwan Correspondence should be addressed to Chien-Hao Liu; cliu82@ntu.edu.tw Received 14 December 2019; Revised 16 March 2020; Accepted 30 June 2020; Published 13 July 2020 Academic Editor: Nigel Zheng Copyright © 2020 Ming-Hwa Hsieh et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. In this research, we proposed a lower-limb soft exoskeleton for providing assistive forces to patients with muscle weakness during the preswing phase of a gait cycle. Whereas conventional soft exoskeletons employ two motors to assist each leg individually, we designed a single motor for actuation. Our design assists hip flexion for light weights and prevents some slip problems that can arise from rotary motors. The actuation mechanism was based on a pulley system that converted the power supplied by the single motor into linear reciprocating motions of a slider. When the single motor rotated, the slider moved linearly, first in one direction and then in the opposite direction. The slider pulled knee braces through cables with an assistive force of 100 N. The actuation was triggered when the system detected that the backward swing of the wearer’s thigh had ended. A prototype was designed, fabricated, and examined with 7 subjects (average age, 24). Subjects were measured while they wore our exoskeleton in power-off and power-on modes. Comparisons proved that wearing the exoskeleton caused a negligible deviation of gait, and that the soft exoskeleton could reduce metabolic cost during walking. The research results are expected to be beneficial for lightweight soft exoskeletons and integration with exosuits that provide assistive forces through the wearer’s entire gait. 1. Introduction stamina, such as walking [11] and driving [12]. Recently, lower-limb exoskeletons have attracted considerable atten- Robotic rigid exoskeletons are commonly used for various tion for providing walking assistance or load-carrying capa- bility through because they can shift the load from the applications, including action assistance, augmentation, and rehabilitation. Rehabilitation-based exoskeletons are usually wearer and pass the load to the ground [13]. However, a dis- mounted on stationary facilities such as treadmills; patients advantage of rigid exoskeletons is that they are heavy and with stroke or injuries wear these exoskeletons for gait thus require undesired metabolic expenditures [14]; addi- retraining or rehabilitation [1–5]. Various portable rigid exo- tionally, they usually impose kinematic constraints when skeletons have been developed; impaired patients wear these the wearer tries to walk [15]. Few rigid exoskeletons can to regain functional abilities. When paralyzed patients wear achieve metabolic reduction for problems such as angle assis- these exoskeletons, they can walk on their feet again [6]. Even tance [16], regardless of whether they are worn for tethered if some elderly patients lack muscle strength, they can wear or stationary activities [17]. exoskeletons to regain their grabbing and gesturing abilities Soft exoskeletons composed of artificial muscles and [7]. In addition, gear-based portable rigid exoskeletons can cable-based actuation mechanisms have become very popu- enhance the strength of healthy wearers for conducting lar in the early twenty-first century because they are light- heavy-weight tasks, such as carrying loads [8] and heavy weight and comfortable. Traditional air-powered artificial lifting [9, 10], or to conduct long-term activities that require muscles require bulky air sources and often lack crucial 2 Applied Bionics and Biomechanics One walking gait cycle actuation properties [18]; cables are widely utilized as actua- tion mechanisms for soft exoskeletons due to their high stiffness. For example, some wearable soft exoskeletons (i.e., exosuits) exploit ribbons [19] or Bowden cables to transmit tensional forces and generate assistive torques on the joints; Leg of interest thus, they can act much like real human muscles. Exosuits Heel Heel Toe Heel can provide effective assistance for hip extension [19, 20], Event strike strike off off Assisted hip flexion, and ankle plantar flexion during toe-off phase Preswing Phase Stance phase Swing phase [20–22] with large savings of energy during walking. A per- Gait 0% 10% 50% 60% 100% son walking on a treadmill at 1.5 m/s, carrying a load equal 30% of his weight, can reduce his metabolic rate by 15% if Figure 1: Illustration of the assistance provided by a lower-limb soft he wears an exosuit [23]. exoskeleton during a walking gait cycle in which assistive forces In a single gait cycle, the leg that swings backward relative were supplied for hip flexion during the preswing phase. to the body can be called the “stance leg” and the leg that swings forward can be called the “swing leg.” Most exosuits hip flexion assistance. The length, width, and height of the provide assistance during the various phases of the stance exoskeleton unit are 202, 165, and 212 mm, respectively. leg and the toe-off phase of the swing leg (Figure 1). However, The exoskeleton unit comprises a start button, an actuation the other phases of the swing leg (e.g., the preswing phase) unit, sensors, a controller, and a battery (Figure 2(c)). The are often neglected. Analyses of these phases are crucial for exoskeleton unit weighs 4.3 kg and contains a motor that patients who lack muscle strength in their lower limbs. can provide a 100 N force at either knee brace. Therefore, the aim of this study was to design a soft exoskel- Figure 3 displays the actuation mechanism of the soft eton to provide hip flexion assistance during walking for exoskeleton. The motor rotates clockwise and moves the wearers who can walk but lack either leg strength or stamina. slider to the left based on the pulley system displayed in The soft exoskeleton should be lightweight and highly com- Figure 3(a) when the actuation is triggered at the preswing pliant with human bodies to avoid imposing kinematic con- phase of the right leg. The linear motions of the slider pull straints on wearers conducting daily activities. Cables with the knee brace of the right leg by the straddle cable and con- spools are commonly used as actuators for soft exoskeletons vert the motor rotation into assistive force applied at the right because they are compact and require little space. However, knee brace. In this case, the straddle cable of the left knee because force transmission relies on the frictions between brace is slacked. For the next gait cycle, the motor must rotate the spools and cables, slip problems occur and cause drifts in a counter-clockwise manner to assist hip flexion for the left in the angular positioning of gaits. Additional space is leg by pulling the straddle cable of the left knee brace. During required for installing force sensors when cables are slacked. either preswing phase, the assistive force is applied at one In this study, a single-motor-based actuation mechanism was knee brace because the system pulls the straddle cable con- designed; this mechanism enabled the proposed lightweight nected to the knee brace, as displayed in Figure 3(b). exoskeleton and prevented the potential slip problems of rotary motors. The actuation mechanism was based on a pul- 3. Modeling and Simulations ley system that converted the power supplied by the single motor into linear, reciprocating slider motions. When the 3.1. Two-Dimensional Kinematic Model of the Human Legs. A two-dimensional (2D) kinematic model of a human leg single motor rotated, the slider moved linearly in one direc- tion to pull the appropriate knee brace with an assistive force (Figure 4) was used to analyze the mechanical power of 100 N through cables. required for hip flexion or extension during a walking gait This paper is organized as follows. Section 2 introduces cycle. The thigh, calf, and foot were modeled as linkages joint the system overview of the soft exoskeleton and its actuation at the ends, and their rotation was limited to the sagittal mechanism. Section 3 presents kinematic analyses and simu- plane. In Figure 4, o, a, b, and c represent the hip joint, knee lations to discuss the conservation of mechanical power in joint, ankle joint, and toe, respectively. The thigh, calf, and the soft exoskeleton. Sections 4 and 5 describe the designs foot lengths were l , l , and l , respectively. The central 1 2 3 and control strategy of the soft exoskeleton, respectively. masses of the thigh, calf, and foot were denoted as m , m , 1 2 Finally, Section 6 reports how the soft exoskeleton was exper- and m , respectively, located on 44% of l from o, 40% of l 3 1 2 imentally examined with several subjects to investigate the from a, and 25% of l from b, respectively. The moments of metabolic costs and the influences on the gaits. The crucial inertia at the central masses of the thigh, calf, and foot were results and conclusions are summarized in Section 7. I , I , and I , respectively. The rotation angles of the thigh, 1 2 3 calf, and foot linkages were θ , θ , and θ with respect to 1 2 3 the horizontal ground. The physical parameters of the 2D 2. System Overview model of the human leg are summarized in Table 1 [24]. Figures 2(a) and 2(b) display the proposed lower-limb soft Based on the average of the ground reaction force and gait exoskeleton comprising a sling strap, a waist belt, knee locomotion data obtained from 20 young people and 20 braces, straddle cables, and an exoskeleton unit. This device adults reported in [25], the mechanical power required for provides assistive forces to the human body during walking hip flexion and extension could be obtained by multiplying by pulling the straddle cables connected to knee braces for the hip rotation moments and the angular velocity of the 212 mm 202 mm Applied Bionics and Biomechanics 3 2 2 2 l + l − a + b Sling strap cable 1 −1 φ = cos : ð2Þ Waist belt 2l l 1 cable Waist belt hook Straddle cable Brace hook Because the direction of the assistive force was in parallel with the straddle cable, the horizontal and vertical compo- Straddle cable sheath Exoskeleton unit nents of the assistive force could be obtained based on φ as Knee brace follows: ψ = − φ − θ ð3Þ hip In the context of the assistive forces in the inverse (a) (b) dynamic analysis, the mechanical power required for hip flexion and extension while wearing the soft exoskeleton is described in the next section. 3.3. Simulation Result. To investigate the influence of the soft exoskeleton on energy conservation, the mechanical power values obtained when the soft exoskeleton was worn and not worn were obtained through the inverse dynamic method. We assumed that our device could provide a con- stant assistive force of 100 N for hip flexion assistance during (c) the preswing phase for 50% to 65% of a gait cycle. As men- Figure 2: (a) Front and (b) back views of the wearable soft tioned in the previous section, the direction of the assistive exoskeleton. (c) Exoskeleton system comprising a start button, an force varied as with the angular position of the hip (Equation actuation unit, sensors, a controller, and a battery. (1)). Figure 6 presents the simulated hip moment, angular velocity of the hip joint, and mechanical power obtained dur- walking gait cycle through the well-known inverse dynamic ing a walking gait cycle. Analytical simulations were con- method. The simulation results are described later. ducted using MATLAB (MathWorks, Massachusetts, United States). The solid and dotted lines represent the cases 3.2. Modeling of the Soft Exoskeleton. Figure 5 displays a 2D in which the soft exoskeleton is worn and not worn. In model of a human leg wearing the proposed soft exoskeleton Figure 6(c), the positive value of power indicates the required in which assistive forces are supplied at the knee braces mechanical power supplied by humans for normal gait through the straddle cables. To study the effects of wearing motions. In other words, the positive area represents the the soft exoskeleton on the mechanical power required for energy consumed by humans while walking. Figure 6(a) dis- hip flexion and extension, the assistive forces were included plays that the magnitude of hip moment required during the in the aforementioned inverse dynamic analysis. preswing phase decreased due to the assistive force provided According to the 2D geometric model, the anchor point by the soft exoskeleton. This result indicated that less hip on the waist had a relative horizontal distance of a and a ver- moment was required for swinging the leg forward. More- tical distance of b. The straddle cable with a length of l cable over, the mechanical power decreased during the preswing was connected the anchor and knee points for assisting hip phase, thus decreasing the energy required from 0.3440 to rotations. The convectional hip rotation angle θ is the hip 0.2873 J/kg per cycle. The assistive force provided by the soft angle between the thigh and the vertical axis and can be exoskeleton could provide hip flexion assistance by conserv- expressed as follows: ing 3.589 J of energy required during walking. θ = θ − 270 . Here, φ represents the angle between the hip 1 tight and the straddle cable and ψ is the angle between the 4. Designs and Assembly of Soft Exoskeleton straddle cable and horizontal axis. The orientation of the straddle cable varied with the hip rotation during walking, 4.1. Pulley System. Most cable-driven actuations of the soft thus indicating that l is a function of θ and can be skeleton tend to exhibit slip problems. Deviations in the hip cable hip rotation positions were observed due to these problems. To expressed as follows: avoid these problems, a slider with reversible linear motions rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi actuated through the pulley system was developed in this pffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 2 2 2 2 −1 l = a + b + l − 2l a + b × cos π − tan − θ : study (Figure 7). All rotating components were installed on cable 1 1 hip an 8 mm thick acrylic board base (Figure 7(a)). A brushless ð1Þ DC motor and two timing belt pulleys were mounted on the rear side of the base, and four pulleys, including guiding With variation in the orientation of the straddle cable, the and driving pulleys, were mounted on the front side of the angle φ varied as the hip rotated and can be expressed as base. The two timing belt pulleys formed a gear train with follows: the gear ratio of 3 to drive the pulley for scrolling a nylon 165 mm 4 Applied Bionics and Biomechanics AF Pulley Motor Slider Load cell AF Knee brace (a) (b) Figure 3: Illustration of the actuation mechanism. (a) When the motor rotates, the slider moves through the pulley system and provides assistive forces for hip flexion by pulling the straddle cables of the knee braces and slacking the straddle cable of the other knee brace. The motor rotates reversely for providing hip flexion assistance to the other leg. (b) During the preswing phase, the assistive force is applied at one knee brace because the motor pulls a straddle cable. Sagittal plane Hip joint 𝜃 m , I 1 1 Extension Flexion Thigh Knee joint m , I 2 2 l Calf m , I 3 3 Ankle joint Foot (a) (b) Figure 4: Two-dimensional kinematic model of a human leg. thread whose both ends were attached to the two sides of the served a crucial role in precisely positioning the slider. slider (Figure 7(b)). When the motor rotated, the timing belt Because the nylon thread exhibited a large tensional stiffness pulley pulled the nylon thread and actuated the slider to value, small assembly misalignments or structure deforma- move in one direction through the driving and guiding pul- tions could easily slack the nylon thread. Although the thread leys. Because the pulley system was operated through the had been wrapped multiple times, slips occurred when the nylon thread and the end of the nylon thread was not fixed nylon thread slacked. Therefore, a pretension mechanism on the driving pulley, friction forces between the nylon was used to maintain the tightness of the nylon threads and thread and the driving pulley were key factors that caused slip was mounted on the slider (Figure 7(b)). To simplify the problems at the nylon thread. To prevent these slip problems, assembly, a torsional spring was incorporated in the slider the driving pulley was wrapped with eight turns of nylon with two ends of the spring tangled with each of its arms thread. The number of turns was limited because a high (Figure 8). The spring was pinned on the slider at its center. number of turns of the threads could increase the frictional The spring could rotate with a small angle such that a forces of the rotating pulleys. Without self-locking effects 1 mm position deviation of the slider was allowed. The spring and a relative low gear ratio of the timing belt pulley, revers- was twisted when the slider was loaded, thus allowing ible motions of the slider were achieved. An advantage of the another 1 mm position deviation. Two screws and nuts were linear slider was that the positioning drift could be eliminated used for each end of the nylon thread to fix both ends firmly, by tracking the positions of the linear sliders by photointer- which are marked as 1a, 1b, 2a, and 2b in Figure 8. Before the rupters, as described later. nylon thread was tightened, the nylon thread was wrapped with one turn on each screw in a clockwise manner on 1a 4.2. Pretension Mechanism. In addition to the friction coeffi- and 2a and in a counter-clockwise manner on 1b and 2b. Then, the nuts on 1a and 2a were screwed to hold and tighten cients and wrapping turns, the tensions in the nylon threads Applied Bionics and Biomechanics 5 Table 1: Physical parameters of the human legs [23]. Thigh Calf Foot l l l 41 cm 41 cm 20 cm 1 2 3 m m m 7.7 kg 3.1 kg 0.8 kg 1 2 3 2 2 2 I I I 1093 kg/cm 406 kg/cm 31 kg/cm 1 2 3 25% of l CM to “o” 44% of l CM to “a” 40% of l CM to “b” 1 2 3 AF during walking. The rest of the strap was wound on the 3D- Anchor point on waist printed connectors. Hip joint Straddle 5. Sensors and Control Strategy cable 5.1. Load Cells. Instead of using commercial load cells, a com- hip pact twin pull-type load cell was designed for the soft exo- cable l skeleton, as illustrated in Figure 10. In Figure 10(a), the load cell connected the heads of the straddle cables and the slider for pulling the knee braces; because the load cell had been designed to be compact, it only occupied a small vol- ume. Finite element simulations were conducted for the x Knee joint designs of the load cell in Comsol commercial software. The simplified quartered model based on two symmetric sur- Figure 5: Two-dimensional modeling of a human leg wearing the faces is presented in Figure 10(b). Because the maximum soft exoskeleton in which the assistive forces were applied at the assistive force was 100 N for the soft exoskeleton, the load cell knee joint by pulling the straddle cable. AF: assistive forces. was designed to measure the maximum assistive force of 200 N with a safety factor of 2. When a straddle cable applied a tensile assistive force of 200 N in the x-direction, the force the nylon thread. Finally, the nuts were screwed on 1b and 2b was applied at the contact point with the x-component of to fix the nylon thread. As the nylon thread undergoes ten- 100 N and y-component of 65.5 N due to the contact angle sion, the friction force between the nuts and the nylon thread of 33.23 between the load cell and head of the straddle cable. tightens the nut further due to the screw and nut mechanism. We assumed the force was exerted on the fillet surface with a Thus, the clamp of the nylon thread does not loosen, even radius of 0.5 mm, which was the surface between the head of after prolonged stress. the straddle cables and load cells. The load cell was made of stainless steel; it was designed to endure a maximum yield 4.3. Anchors and Connectors. We used straddle cables to stress no greater than 200 MPa. The simulation results are transmit assistive forces from our device to the knee braces shown in Figures 10(c) and 10(d). and then to the legs of the exosuit wearer. The connections The maximum stress was 125 MPa and was 37.5% between the cable and the knee braces were designed to be smaller than maximum yield stress, thus indicating that the easily adjustable for different wearers. Because the straddle load cell could withstand a force of 20 kgw. The maximum cables lacked flexibility and adjustability and could not be strain occurred on the upper surface of the load cell at a dis- tied with a knot on the knee braces, Velcro was used to tance of a 14 mm from the center of the load cell in the lon- adhere the straddle cables to the knee braces for ensuring gitudinal direction. This phenomenon caused a small reliable attachments. elongation of 2.04 μm, and a fine resolution could not be Figure 9(a) shows the anchor of the soft exoskeleton attained for the general metallic foil-type strain gages. There- composed of the slot for holding the cable sheath and the fore, strain gages with a large gage factor (GF) as that of open slit for attachment to the waist belt. Since the anchor semiconductor-type strain gages should be used for our attached to the waist close to the hip, walking might cause device. Multiple strain gages (Kyowa: KSN-2-120-E3-16) discomfort to the pelvic region; to maximize comfort, the with a GF of −105 were implemented using two quarter- 3D-printed anchor had a support wing filled with foam that bridge Wheatstone bridges for the twin load cell. The load distributed pressure on the hip. Figure 9(a) shows the 3D- cell was fabricated and calibrated experimentally at a mea- printed connector composed of an M4 screw, an M4 nut, surement range of 0 N–100 N and a resolution of 1.76 N and spacers. By threading the straddle cables through the (Figure 10(e)). Although the load cell exhibited a fine resolu- spacers and clamping it with the screw and the nut, we fixed tion and good linearity in the force measurements, the drift the 3D-printed connector tightly with the straddle cables. effect due to the strain gages and the increase in the temper- Then, the Velcro strap on the knee brace was wound with a ature of the circuit were not negligible. Figures 10(f) and buckle to ensure that the Velcro strap would remain fastened 10(g) display the drift effects of the left and right sides of 6 Applied Bionics and Biomechanics (a) 0.5 −0.5 0 20 40 60 80 100 7.5 5 (b) 2.5 −2.5 −5 0 20 40 60 80 100 Insufficient energy (c) Surplus energy −2 0 20 40 60 80 100 Gait cycle (%) 100 N constant force is applied Figure 6: Simulated (a) hip moment, (b) angular velocity of the hip joint, and (c) mechanical power during a walking gait cycle. Here, the solid and dotted lines represent the cases in which the soft exoskeleton is not worn and worn, respectively. A constant assistive force was applied during the preswing phase for conserving power. X = 0 Photo interrupter s Pretension mechanism Linear slider Stroke Guiding pulleys Driving pulley Timing belt pulley Motor ⁎⁎ Tensioning pulley (a) (b) Figure 7: Illustration of the power transmission based on the slider and the pulley system in the (a) top and (b) side views. the load cell, in which the measured force varied and satu- phase. Moreover, a linear relation existed between the length rated as time increased. Therefore, the measured data of load of cable l and the hip rotation angle θ , which is insen- cable hip cells were used as trigger signals in our preliminary work sitive to the changes in the geometric parameters of l , a, because accurate values of the assistive forces could not be and b, (Figure 4). Equation (1) simplified the linearization provided. method with a small variation in the hip rotation angle: 5.2. Coordinate Mapping. Because the actuation mechanism l ≈ 0:1732θ + c ð4Þ cable hip 1, converted the motor rotations to the linear motions of the slider, the coordinates of the hip angular positions were mapped to linear positions set on the sliders. The hip rotation where c is a constant based on the thigh length. By defining ° ° angles were in the range of −10 to 10 during the preswing c = l at θ =0, the mapped coordinate of the slider X 1 cable hip s Power (W/kg) (rpm) M hip (Nm/kg) Applied Bionics and Biomechanics 7 Clamp for nylon thread photointerrupters were mounted along the stroke with the mapped positions of X = −4:5, −2,−1, 0, 1, 2, and 4:5cm to track the positions of the linear motions of the slider. 1a Figure 12 displays the flowchart of the control strategy that 1b 2b 2a was realized for different stages—idle, wearer-driven pulling, actuation, and brake stages. They were determined using the motor rotational speed measured from the embedded speed sensor of the motor. If the motor rotation was higher than Nylon thread Torsional spring Turning point 800 rpm, then brakes were applied automatically and deceler- ated until the rotation speed was lower than 50 rpm. The Figure 8: Diagram of the pretension mechanism and fixation of motor rotational speed was zero while walking. Moreover, nylon threads mounted on the slider. the soft exoskeleton was in the idle stage, and the slider was at the position of X = −1or1cm at which the straddle cables of both legs were slacked. The load cells were enabled to (a) monitor the tensile forces of the straddle cables connected to the knee braces. During the end period of the backward swinging of legs, the slider was moved away from the posi- tions of X =±1 cm and the soft exoskeleton entered the wearer-driven pulling mode. At the beginning of the pre- swing phase, one side of the load cell detected a high decrease in the measured force when the leg swung forward. The actu- ation mechanism was triggered when the motor begin to rotate and thus the slider began to move linearly to provide hip flexion assistance; this period can be known as the actu- (b) ation period. When the slider passed the position of X =0 cm, the soft exoskeleton entered the brake stage and the motor rotated in the opposite direction to decelerate the lin- ear motion of the slider until the slider stopped at the posi- tion of X =±1cm. Then, the slider remained at the position of X =±1cm and the soft exoskeleton entered the idle stage until the next actuations. The photointerrupters at X = −4:5, −2, 2, and 4:5cm were designed to ensure that the slider returns to the position of X =0 cm in our original Figure 9: (a) Anchors and (b) connectors mounted on the waist. design, which is not conducted in the present study. 5.4. Actuation Trajectory. The motion trajectory of the linear was defined as the origin at which X =0 cm when the slider slider while wearing the soft exoskeleton is shown in was at the midpoint of the stroke at length of 90 mm Figure 11 based on the control strategy. At the beginning of the preswing phase, that is, after the heel of the right leg (Figure 7(b)). Based on the above condition, the formula of X can be presented as follows: strikes the ground (t =0 s), the actuation mechanism was activated due to a rapid decrease in the force measured force X =0:1732θ ð5Þ using the load cell and an assistive force was provided at the s hip: right knee by pulling the straddle cable through the linear movement of the slider by the counter-clockwise rotation of The thigh motions were projected to the mapped coordi- the motor. When the slider passed the position of X =0 cm nates by substituting θ into Equation (5). Here, the tar- s hip ° , the soft exoskeleton entered the brake stage and the motor geted hip extension or flexion angles in the range of −10 to rotated in a clockwise manner to decelerate the slider until 10 were mapped to the X range of −2 to 2 cm for the the slider stopped at the position of X = −1cm. Subse- mapped coordinates. We assumed that the walking gait of a quently, the soft exoskeleton entered the idle stage and the normal individual was bilaterally symmetrical in terms of slider remained at the same position of X = −1cm at which the phase and motion. Therefore, the slider motion projec- the straddle cables of both legs were slacked. Similarly, at tion of the left thigh X was a mirrored version of the right the end period of the backward swinging of the left leg, the thigh X with a 50% phase lag during gait cycle. The slider slider was pulled passively and moved away from the position motion projections for both thighs when the soft exoskeleton of X = −1cm. Then, at the beginning of the preswing phase was not worn are displayed in Figure 11. In this figure, the of the left leg, the actuation mechanism was triggered and the dotted line represents the right leg and the dotted–dashed motor rotated in a clockwise manner to provide an assistive line represents the left leg. force at the left knee for hip flexion assistance. At this point, 5.3. Control Strategy. A position-based control strategy was the slider moved to the position of X =1cm due to the used for the linear actuation mechanism in this study. Seven motor and pulled the straddle cable connected to the left R3.65 8 Applied Bionics and Biomechanics 1.25×10 (b) −2 ×10 ×10 1.5 35 0.5 1.2 𝛷 6 0.8 −3 ×10 0.6 0.4 0.6 0.8 0.2 12 1.2 −2 ×10 5.77×10 Position of maximum strain −4 6.31×10 −4 on upper surface ×10 −2 6 (a) 1.5 ×10 0.5 (d) −3 ×10 3 0.6 Fillet 0.8 1.2 Cross section for force applying −2 ×10 (fix boundary) z −7 3.95×10 (c) (e) (f ) (g) Figure 10: FEM simulations of the load cell. (a) Geometrical dimension (unit: mm). (b) Simplified quartered model. (c) Principal stress distributions. (d) Principal strain distributions. (e) Photograph of the load cell. (f, g) Drift effects of the left and right sides of the load cell. AF AF knee. When the slider passed the position of X =0 cm, the soft exoskeleton entered the brake stage and the motor rotated in a counter-clockwise manner to decelerate the Driven passive slider until the slider reached the position of X =1cm. Brake Heel strike Then, the slider remained at the position of X =1 cm Right leg and the soft exoskeleton entered the idle stage until next gait cycles. Idle −2 6. Experimental Results and Discussions Actuation −4 6.1. Metabolic Energy Conservation. To evaluate the capabil- Left leg −6 ities of conserving energy when a wearer of the soft exoskel- 0 0.2 0.4 0.6 0.8 1 eton walks, metabolic tests were conducted using a treadmill Time (s) machine (Figure 13(a)). Seven participants with ages ranging from 23 to 24 years, height ranging from 171 to 180 cm s (average = 174 cm, standard deviation ðSDÞ =3:16 cm), and weight ranging from 67 to 72 kg (average = 70 kg, SD = 1:91 Figure 11: Illustrations of the motion projections of the mapped kg) participated in the tests. The treadmill was setup at a con- coordinate for the cases in which the soft exoskeleton is worn and stant walking speed of 4.3 km/h. The metabolic rates of the not worn while walking. The assistive forces were provided during subjects were measured and recorded using a portable the preswing phases for each leg. pulmonary gas exchange measurement system (K4b2, COSMED s.r.l., Rome, Italy) during the test. The tests included four scenarios: (a) wearing the knee braces only X (cm) 6 12 Applied Bionics and Biomechanics 9 Idle Reading Start Open load cell Big surge ? load cell value Photo T F interrupter X = ±1 input Motor running Speed input output Speed>800 Photo rpm interrupter X = 0 input s T Actuation Brake Brake start Close Motor running Speed< load cell stopping 50 rpm Figure 12: Flowchart of the control algorithm. Pulmonary gas exchange measurement system Exoskeleton −10 −20 Treadmill −30 Loaded Power off Power on (a) (b) Figure 13: (a) Photograph of the metabolic tests conducted while walking on a treadmill. (b) The measured statistical distributions of the metabolic rate reduction for the cases in which loads are carried, power is switched off, and power is switched on. The error bar indicated one standard deviation. The metabolic rates were normalized to the case in which only the knee braces were worn (i.e., without wearing the soft exoskeleton). (i.e., free walking without wearing the soft exoskeleton), (b) worn when it is switched off, and worn when it is switched wearing the soft exoskeleton when the straddle cables were on. The metabolic rate reduction is expressed as follows: loosened (i.e., carrying loads), (c) wearing the soft exoskele- ton when it is switched off, and (d) wearing the soft exoskel- MR − MR m f eton when it is switched on. For the scenarios (c) and (d), a : ð6Þ MR 2 kgw pretensioned force was applied while wearing the soft exoskeleton. Each test lasted for 7 min. Because our device was a prototype that had not been optimized for weight Here, MR is the metabolic rate when the soft exoskele- and comfort, we focused on the effects of the assistance and ton is worn for three cases and MR is the metabolic rate ignored the influence of carrying loads. To ensure that the obtained without wearing the soft exoskeleton. The negative soft exoskeleton could be triggered appropriately and could values represented that energy was conserved due to the actuate correctly, the straddle cables were pretensioned with assistance provided by the exoskeleton, and the positive a 2 kgw force while a subject was wearing the soft exoskele- values presented that energy consumption occurred when ton. Moreover, the anchors on the waist were adjusted such loads were carried. These results demonstrated that the that the cables were in parallel with the sagittal plane. metabolic rate could be reduced further due to the assistive Figure 13(b) displays the measured statistical distribu- forces provided by our device while walking, thus indicating tions of the metabolic rate reduction for the cases in which that the proposed soft exoskeleton can conserve energy when the soft exoskeleton was worn with loosened straddle cables, the wearer walks. The deviations were due to the Metabolic rate reduction (%) 10 Applied Bionics and Biomechanics (a) Accelerated −10 −20 0 20 40 60 80 100 Phase delay Gait cycle (%) Power on Power off Free (b) Power on Power off Free 30 30 30 20 20 20 10 10 10 0 0 0 𝜃 𝜃 𝜃 −10 −10 −10 −20 −20 −20 −30 −30 −30 0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100 Gait cycle (%) Gait cycle (%) Gait cycle (%) (c) (d) (e) Figure 14: (a) Photograph of the gait motion measured using the video camera. (b) Averaged gait motions for the cases in which the soft exoskeleton is worn with its power switched on, is worn with its power switched off, and is not worn. An assistive force was provided by the soft exoskeleton during the preswing phase. (c, d, e)All gait motions for each scenario. track the positions of three markers. The angle formed by unfamiliarity in using the exoskeleton or lack of soft exoskel- eton training. the markers were measured, transformed, detrended, and finally separated through every two heel strikes in one gait 6.2. Impacts on Gaits. To investigate the effects on gait cycle. motions when the soft exoskeleton is worn, gait measure- Figure 14 displays the measured gait motions of the sub- jects. The solid lines represent that the soft exoskeleton was ments were conducted for the same subjects. Subjects wore a soft exoskeleton and walked along a straight line with three worn and was powered on, and the dashed lines represent markers attached at their shoulders, hips, and knees. More- that the soft exoskeleton was worn and was powered off. over, the angular rotation positions of the hip and knees were The dotted lines represent that only the knee braces were recorded using a video camera located at a distance of 5.3 m worn (i.e., free walking). The three curves presented in in front of the straight line (Figure 14(a)). The experiments Figure 14(b) were obtained through fitting the curves of all were conducted for the participants under three scenarios—- the gait motions for the scenarios presented in wearing the soft exoskeleton when it is switched on, wearing Figures 14(c)–14(e). A phase lag from 35% to 56% is dis- the soft exoskeleton when it is switched off, and wearing knee played in Figure 14(b); subjects attain the maximum hip braces only (i.e., free walking without wearing the soft exo- extension while wearing the device when it is switched off rel- skeleton). In the first two scenarios, a 2 kgw pretensioned ative to the free walking case. This behavior was observed force was applied while when the soft exoskeleton was worn. because that the slider remained at the last position after pull- The gait patterns were recorded at a frame rate of 30 fps. Each ing and could not return to the central point when the device frame was analyzed with image processing techniques to was switched off. When the soft exoskeleton was switched on, (deg.) hip (deg.) hip (deg.) hip (deg.) hip Applied Bionics and Biomechanics 11 the damping effect was minimized because the slider could References move and return to the central position after being pulled [1] C. H. Wu, H. F. Mao, J. S. Hu, T. Y. Wang, Y. J. Tsai, and W. L. by the subject. Therefore, the phase lag of the maximum Hsu, “The effects of gait training using powered lower limb hip extension was eliminated when the soft exoskeleton was exoskeleton robot on individuals with complete spinal cord switched. Moreover, hip rotation accelerated in the preswing injury,” Journal of NeuroEngineering and Rehabilitation, phase from 35% to 56% due to the hip flexion assistance pro- vol. 15, no. 1, p. 14, 2018. vided by the soft exoskeleton (Figure 14(b)). Because no [2] S. Jezernik, G. Colombo, T. Keller, H. Frueh, and M. Morari, obvious changes were observed in the gait, the proposed soft “Robotic orthosis Lokomat: a rehabilitation and Research exoskeleton is concluded to have at most a slight influence on Tool,” Neuromodulation, vol. 6, no. 2, pp. 108–115, 2003. the gait motions. [3] K. S. Banala, S. K. Agrawal, and J. P. Scholz, “Active leg exo- skeleton (ALEX) for gait rehabilitation of motor-impaired patients,” in 2007 IEEE 10th International Conference on Reha- 7. Conclusions and Future Work bilitation Robotics, Noordwijk, Netherlands, June 2007. A compact lower-limb soft exoskeleton was presented in this [4] J. Bae, S. M. M. D. Rossi, K. O'Donnell et al., “A soft exosuit for study for providing hip flexion assistance to walking wearers patients with stroke: feasibility study with a mobile off-board with low muscle strength. A lightweight exoskeleton was actuation unit,” in 2015 IEEE International Conference on Rehabilitation Robotics (ICORR),, Singapore, Singapore, designed to minimize slip problems; the soft exoskeleton August 2015. was designed to actuate both legs through a single actuator by converting the motor rotations into linear reciprocating [5] J. L. Contreras-Vidal, N. A. Bhagat, J. Brantley et al., “Powered exoskeletons for bipedal locomotion after spinal cord injury,” motions of the slider through a pulley system. According to Journal of Neural Engineering, vol. 13, no. 3, p. 031001, 2016. the simulation results, the exoskeleton was capable of con- [6] A. Esquenazi, M. Talaty, A. Packel, and M. Saulino, “The serving the energy required during the preswing phase of a rewalk powered exoskeleton to restore ambulatory function gait cycle. A prototype was fabricated, assembled, and exper- to individuals with thoracic-level motor-complete spinal cord imentally examined through metabolic rate tests. The results injury,” American Journal of Physical Medicine & Rehabilita- revealed that our device could reduce the metabolic cost of tion, vol. 91, no. 11, pp. 911–921, 2012. walking and exerted at most a slight influence on gait [7] P. Polygerinos, K. C. Galloway, S. Sanan, H. Maxwell, and C. J. motions, thus indicating that the proposed soft exoskeleton Walsh, “EMG controlled soft robotic glove for assistance dur- could conserve energy during hip flexions. ing activities of daily living,” in 2015 IEEE International Con- Some future studies are under investigation. Because the ference on Rehabilitation Robotics (ICORR), Singapore, actuation speed of straddle cables is limited, the current actu- Singapore, August 2015. ation mechanism cannot be used to attain a high walking [8] C. J. Walsh, K. Endo, and H. Herr, “A quasi-passive leg exo- speed. Moreover, the load cell has a significant drift effect skeleton for load-carrying augmentation,” International Jour- because of the temperature rise in the strain gage and nal of Humanoid Robotics, vol. 4, no. 3, pp. 487–506, 2011. circuitry. This problem can be overcome by finely soldered [9] W. Micheal and W. Michael, “Lower extremity exoskeleton circuitry and temperature compensations. Moreover, the reduces back forces in lifting,” in ASME 2009 Dynamic Systems measured load information can be utilized for feedback con- and Control Conference, Hollywood, January 2009. trol. Because the soft exoskeleton was designed to provide [10] K. Yamamoto, M. Ishii, K. Hyodo, T. Toshimitsu, and assistance for the preswing phase of a gait cycle, our device T. Matsuo, “Development of power assisting suit,” JSME Inter- could be integrated with other exosuits to provide assistance national Journal Series C Mechanical Systems, Machine Ele- for entire gait cycles in the future. The proposed single ments and Manufacturing, vol. 46, no. 3, pp. 923–930, 2003. actuator-based actuation mechanism is expected to be [11] Y. Hiki, Z. Sugawara, and J. Ashihara, “Walking assist device,” beneficial for realizing a compact and lightweight soft p. B2, 2013, U. S. Patent 8603016. exoskeleton. [12] B. L. Shields, J. A. Main, S. W. Peterson, and A. M. Strauss, “An anthropomorphic hand exoskeleton to prevent astronaut hand fatigue during extravehicular activities,” IEEE Transactions on Data Availability Systems, Man, and Cybernetics - Part A: Systems and Humans, vol. 27, no. 5, pp. 668–673, 1997. The measured data used to support the findings of this study are available from the corresponding author upon request. [13] A. B. Zoss, H. Kazerooni, and A. Chu, “Biomechanical design of the Berkeley lower extremity exoskeleton (BLEEX),” IEEE/ASME Transactions on Mechatronics, vol. 11, no. 2, Conflicts of Interest pp. 128–138, 2006. [14] G. J. Bastien, P. A. Willems, B. Schepens, and N. C. Heglund, The authors declare no conflicts of interest. “Effect of load and speed on the energetic cost of human walk- ing,” European Journal of Applied Physiology, vol. 94, no. 1-2, pp. 76–83, 2005. Acknowledgments [15] A. Schiele and F. C. T. van der Helm, “Influence of attachment This work was supported by the Ministry of Science and pressure and kinematic configuration on pHRI with wearable Technology, Taiwan, under grant numbers 106-2221-E- robots,” Applied Bionics and Biomechanics, vol. 6, no. 2, 002-133-MY2 and 108-2221-E-002-165. pp. 157–173, 2009. 12 Applied Bionics and Biomechanics [16] L. M. Mooney, E. J. Rouse, and H. M. Herr, “Autonomous exo- skeleton reduces metabolic cost of human walking during load carriage,” Journal of NeuroEngineering and Rehabilitation, vol. 11, no. 1, p. 80, 2014. [17] Y. Ding, I. Galiana, A. Asbeck, B. Quinlivan, S. M. M. De Rossi, and C. J. Walsh, “Multi-joint actuation platform for lower extremity soft exosuits,” 2014 IEEE International Conference on Robotics and Automation (ICRA), 2014, Hong Kong, China, June 2014, 2014. [18] Y.-L. Park, B.-r. Chen, D. Young et al., “Bio-inspired active soft orthotic device for ankle foot pathologies,” in 2011 IEEE/RSJ International Conference on Intelligent Robots and Systems, San Francisco, CA, USA, September 2011. [19] A. T. Asbeck, K. Schmidt, and C. J. Walsh, “Soft exosuit for hip assistance,” Robotics and Autonomous Systems, vol. 73, pp. 102–110, 2015. [20] A. T. Asbeck, K. Schmidt, I. Galiana, D. Wagner, and C. J. Walsh, “Multi-joint soft exosuit for gait assistance,” in 2015 IEEE International Conference on Robotics and Automation (ICRA),, Seattle, WA, USA, May 2015. [21] A. T. Asbeck, R. J. Dyer, A. F. Larusson, and C. J. Walsh, “Bio- logically-inspired soft exosuit,” in 2013 IEEE 13th Interna- tional Conference on Rehabilitation Robotics (ICORR),, Seattle, WA, USA, June 2013. [22] A. T. Asbeck, S. M. M. De Rossi, K. G. Holt, and C. J. Walsh, “A biologically inspired soft exosuit for walking assistance,” The International Journal of Robotics Research, vol. 34, no. 6, pp. 744–762, 2015. [23] F. A. Panizzolo, I. Galiana, A. T. Asbeck et al., “A biologically- inspired multi-joint soft exosuit that can reduce the energy cost of loaded walking,” Journal of NeuroEngineering and Rehabilitation, vol. 13, no. 1, p. 43, 2016. [24] C. E. Clause, Weight, volume, and center of mass of segments of the human body, Antioch College Yellow Springs, Ohio, 1969. [25] G. Bovi, M. Rabuffetti, P. Mazzoleni, and M. Ferrarin, “A multiple-task gait analysis approach: kinematic, kinetic and EMG reference data for healthy young and adult subjects,” Gait & Posture, vol. 33, no. 1, pp. 6–13, 2011.

Journal

Applied Bionics and BiomechanicsHindawi Publishing Corporation

Published: Jul 13, 2020

References