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

Learn More →

Method, Design, and Evaluation of an Exoskeleton for Lifting a Load In Situ

Method, Design, and Evaluation of an Exoskeleton for Lifting a Load In Situ Hindawi Applied Bionics and Biomechanics Volume 2021, Article ID 5513013, 12 pages https://doi.org/10.1155/2021/5513013 Research Article Method, Design, and Evaluation of an Exoskeleton for Lifting a Load In Situ Xin Li , Weihao Li , and Qiang Li School of Mechanical and Materials Engineering, North China University of Technology, Beijing 100144, China Correspondence should be addressed to Xin Li; lixin2020@ncut.edu.cn Received 15 February 2021; Revised 22 April 2021; Accepted 7 May 2021; Published 25 May 2021 Academic Editor: Andrea Cereatti Copyright © 2021 Xin Li 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. Due to the unclear application scenarios and force analysis of exoskeletons, there exists a research gap in exoskeleton design. This paper presents a design method and realization of an exoskeleton for a specific scenario of lifting a load in situ. Firstly, the lifting motion process and its data were collected based on a 3-D motion capture system and dynamometer treadmill system. Then, the variations of the torque and motion of each joint were obtained from the data analysis, based on which an active assistance mode for upper limbs and a passive assistance mode for lower limbs were demonstrated. In this design, the hydraulic cylinder for shoulder assistance, the motor for elbow assistance, and the spring for lower limb assistance were calculated and selected according to the motion and torque of each joint. Finally, subjective and objective methods were used to evaluate the exoskeleton based on the results of five test participants, and the median oxygen consumption of the whole test by lifting a load ten times with the assistance was found to be reduced by 9.45% as compared with that in the absence of the exoskeleton. 1. Introduction conduct lifting. Generally, active assistance and passive assistance [8] are two common driving modes, and active The rapid development of internet technology has not only assistance modes can be further divided into hydraulic [9], introduced convenience to daily life but has also promoted motor-driven [10, 11], and pneumatic [12, 13] modes. The the expansion of the logistics industry. Robots have been power systems of hydraulic or pneumatic driving modes intro- applied in some logistics tasks, such as handling, grabbing, duce additional load to the body due to their large volume; in and placing items with regular shapes; however, robots can- contrast, a motor can be placed on the back of the body. not replace human beings for the handling of goods with Among the limb assistance components, assistance for upper limbs (shoulder [14], elbow [15], and wrist [16] joints), the uncertain shapes, sizes, or weights. There are nearly 50 mil- lion logistics employees in China, and many (<40 years old) waist joint [17], and lower limbs (hip [18], knee [19], and who work under high-intensity conditions without limb ankle [20] joints) has been developed. Moreover, some protection have suffered from tenosynovitis, lumbar disc exoskeletons, called exosuit, have no rigid frame [21]. An exo- herniation, or other diseases. Therefore, it is necessary to pro- suit is driven by a motor and a Bowden cable that is fixed on vide protection to ensure the health of the limbs and joints of the end of the motor, and force can be transferred to any limb these workers. joints by the Bowden cable [22]. From the perspective of con- An exoskeleton is a functional device that attaches to the trol methods, electromyography (EMG) signals [23] or end human body to assist specific joints to protect or strengthen force/torque detection [24] are usually regarded as the control the body. According to the different application scenarios of input. Regarding exoskeleton evaluation, the subjective feeling products, the modes of exoskeletons can be generally of wearers and the results of an objective test have been illus- categorized into rehabilitation [1–5], industry [6], and military trated to evaluate a passive exoskeleton [25]. [7] applications; for industry and military applications in par- As is evident from the preceding review, there are numer- ticular, the exoskeleton supplements normal human power to ous structural, driving, and control modes that can be adopted 2 Applied Bionics and Biomechanics in exoskeleton design. However, exoskeletons are not general- motion, and the assistance performance must be included as ized devices; the ignorance of the influence of the working design inputs. The design and analysis process is illustrated in scenario or working load on human limbs will lead to the Figure 1. inapplicability of exoskeleton design. For example, rehabilita- According to the figure, the design process of the exoskel- tion exoskeletons are usually oriented to patients with physical eton can be divided into the following five steps: the input of disabilities, who rely on the exoskeleton to provide fixed the working condition, motion capture, data analysis, system movements to perform rehabilitation and physical therapy design, and test evaluation. First, the input of the working on the limbs. Therefore, the emphasis of rehabilitation exo- condition requires the analysis of exoskeleton application skeletons is usually placed on the movement of specificjoints. scenarios, such as the working environment, load profile, The portability of wear and weight are two aspects of and human characteristics, based on which misunderstand- exoskeleton product. In contrast, for industrial or logistics ings of exoskeleton design can be avoided. Second, the torque applications, the target is normal people with a certain labor and motion of the joints should be tested based on 3-D intensity. In this case, comfortable wear is almost as important motion capture equipment, from which the values of the as the effectiveness of assistance; otherwise, unfriendly human- torques, angles, and muscle activity of human limbs can be exoskeleton interaction would directly affect the wearer’swork obtained. Third, the joint torque data can guide the maxi- efficiency. Moreover, the design of exoskeletons requires mum capacity and selection of joints for exoskeleton design, analysis to determine which joints or limbs require assistance and the data of joint motion can be used to determine the based on the application scenario, which has been omitted in speed and motion range of limbs. Fourth, the energy and traditional research. drive systems are designed based on the joint torque data There has been less research on exoskeletons for use in and assistance efficiency. Additionally, the structure and industry or logistics applications than on exoskeletons for mechanical system are designed based on the joint motion use in rehabilitation. Li et al. [26] presented an active dual- data. Then, the exoskeleton prototype system is assembled arm exoskeleton that can be adapted to various environments based on the structure and electrical system. Finally, the by adjusting the force and impedance adaptation, such as results of objective tests and subjective feelings about the lifting loads or rehabilitation training. Yu et al. [27] illus- exoskeleton prototype can be evaluated. In particular, the trated an upper-limb exoskeleton for refractory construction, assistance efficiency of the prototype should be compared which can be used to reduce the physical fatigue of operators to the design values, based on which the system design can resulting from long hours of working with heavy loads; how- be optimized. ever, this active exoskeleton has the deficiencies of a greater self-weight and unfitness for walking. Koopman et al. [28] 3. Motion Capture and Data Analysis of Lifting presented a light and convenient exoskeleton for lumbar Load In Situ protection in logistics work. Dinh et al. [21] illustrated an exosuit that reduces the muscular effort required to lift 1 kg 3.1. Experimental Scenario. The scenario considered in this by 48.3%. Picchiotti et al. [17] compared two commercially research was a logistics sorting operator lifting a load from available postural assist exoskeletons and reported that there the ground to a certain height in a fixed operation area. The is no significant biomechanical benefit regarding the joint load mass was set as 20 kg, and the lifting heights were, flexion angles and moment arms for lifting a given load. As respectively, set as 1 m and 1.5 m. To obtain clear and com- is evident, although passive exoskeletons or exosuits are plete data on the lifting process, a 3-D motion capture system characterized by improved human-machine interaction, they (Phasespace, Impulse X2E) was adopted to collect data on the are not suitable for large loads. Consequently, it is important lifting movement of the whole body and each joint; 36 to consider the application scenario and human-machine marker points were stuck on the body of the wearer, and 10 interaction as the design bases. cameras captured the motion. Additionally, a dynamometer This study presents a complete method of exoskeleton treadmill system (Bertec FIT, FITITC-11-20L) was used to design that is oriented to packing and lifting in logistics work. obtain the plantar force (see Figures 2 and 3). Then, Open- The remainder of this research is organized as follows. Section Sim (V4.0) software was used to collect data on the motion 2 presents the design and analysis process of the exoskeleton and plantar force, based on which the motion of the limbs under the suggested scenario. Section 3 illustrates the motion and the torque of the joints of the test participants were capture and data analysis regarding the movement of entire obtained. limbs, based on which the kinematics and torque of joints are obtained. Section 4 reports the calculation and design of 3.2. Analysis of Experimental Data. Figure 3 depicts the the exoskeleton, which consists of elbow, shoulder, and lower experimental scenario of the lifting test. Data were collected limb assistance, as well as an electrical system. Section 5 illus- from the 3-D motion capture system and dynamometer trates the prototype and experiments, the results of which treadmill system to determine the motion of the limbs and verify the effectiveness of the design process. the torque of the joints of the participant, after which the data were imported into OpenSim software to restore the partici- pant’s motion and determine the motion variation process, as 2. Exoskeleton Design and Analysis Process shown in Figure 4. This process was primarily conducted to As mentioned in the previous section, exoskeletons are not uni- verify the effectiveness of the data collection of the 3-D versal products; thus, the working environment, the wearer’s motion capture system and to provide support for the inverse Applied Bionics and Biomechanics 3 Maximum Energy Subjective Joint capacity Environment and evaluation torque Selected drive joints Assisting Working system Assisting Oxygen Human Motion Exoskeleten efficiency condition effect consumption Structure Speed Joint and Load motion EMG data mechanical Range system Input of working condition Motion capture Data analysis System design Test and evaluation Figure 1: The design and analysis process of an exoskeleton. kinematics solution of the torques of the joints in the subse- quent step. Figure 5 reveals the height variation of the lifted load collected by the 3-D motion capture system, which meets Data acquisition the design requirements. Figure 6 presents the variations of 3-D motion capture camera and timing the angles of the hip, knee, ankle, shoulder, and elbow joints system in the extension/flexion degree of freedom (DOF) when lift- Marker point ing the load. The first lifting phase was from the ground to a height of 1 m, and the second lifting phase was from the Load height of 1 m to the height of 1.5 m. As shown in Figure 6, Dyanmometer Computer during the first lifting phase, the lower limbs of the partici- treadmill pant gradually changed from the bending state to the upright ° ° state; specifically, the hip joint changed from 90 to -10 , the Figure 2: Schematic diagram of the lifting test. ° ° knee joint changed from -130 to 0 , and the ankle joint chan- ° ° ged from 32 to -7 . During the second lifting phase, there was almost no change in any joint of the lower limbs. Regarding the upper limbs, the angle of the shoulder joint was 90 when the participant grabbed the load on the ground; it was reduced to 50 after the first lifting phase, and it finally increased to 75 during the second lifting phase. The angle 3-D motion capture camera of the elbow joint changed little during the first lifting phase ° ° and increased from 10 to 55 during the second lifting phase. Additionally, the rotational speeds of the limb joints were Marker point obtained based on Figure 6. In the first lifting phase, the average speeds of the hip, knee, and ankle joints were found ° ° ° to be 75 /s, 76.47 /s, and 18.82 /s, respectively. Regarding the Dynamometer upper limbs, the rotational speeds of the elbow and shoulder treadmill joints in the second lifting phase presented faster variations than in the first lifting phase, and the maximum rotational ° ° speeds were about 40.9 /s and 41.7 /s, respectively. These kine- Load matic values of the limbs and joints provide the basis of the structural and system design of the exoskeleton system. As lifting in situ is primarily completed in the sagittal plane, the extension/flexion DOF of each joint provides more Figure 3: Experiment of lifting test. support for lifting. As shown in Figure 7, the torque value of the flexion DOF was much higher than that of the side-up and rotation DOFs. Therefore, this paper focuses on the analysis of the extension/flexion DOF of each joint. Figure 8 presents the torque variations of the hip, knee, ankle, shoulder, and elbow joints in the extension/flexion DOF during the lifting process. The torque amplitudes of the joints in the lower limb were found to be higher than those of the joints in the upper limb. In addition, the torque Figure 4: The motion animation of the participant restored by values of the upper-limb joints are positive because the OpenSim software. rotational direction of torque acting on the elbow and shoul- der is clockwise; specifically, the latissimus dorsi behind the Height (mm) 4 Applied Bionics and Biomechanics 1600 50 1200 nd 2 liing height st 1 liing height 600 –50 01 2 3 4 4.5 –100 0 1 2 3 4 4.5 Time (s) Time (s) Figure 5: The height variation of the lifted load. Hip Shoulder Knee Elbow Ankle Figure 8: Torque variations of each joint during lifting. shoulder is mainly responsible for its flexion, and the bicipital muscle on the upper limb is mainly responsible for elbow flexion. Regarding the lower limbs, the direction of DOFs of –50 the hip and ankle joints is extension during the entire lifting process, and the direction of torque driven by the corre- sponding muscle to the joint is counterclockwise; thus, the –100 st Holding 1 liing values of their torques are negative. Regarding the knee, phase nd phase 2 liing phase torque assists the extension DOF of the joint in the first lift- –150 ing phase, the direction of which is clockwise, so the value 01 2 3 4 4.5 is positive. Subsequently, the value changes to a negative Time (s) value because the direction of DOF changes to flexion, which Hip Shoulder changes the direction of the joint torque to counterclockwise. Knee Elbow Regarding the torque amplitude, the value of the hip joint Ankle ranged from -95 to -70 Nm, that of the knee joint changed from 40 to -70 Nm, and that of the ankle joint ranged from Figure 6: Variations of joints’ angles during lifting. -45 to -10 Nm throughout the entire lifting phase. During the first lifting phase, the torque of the shoulder joint was found to increase from 17 to 30 Nm, and that of the elbow joint increased from 0 to 10 Nm. During the second lifting phase, the torque of the shoulder joint was found to continu- ally increase from 30 Nm to nearly 50 Nm, while the torque of the elbow joint increased from 10 to 18 Nm and then decreased gradually after reaching the peak; this occurred because the load was closer to the participant, and the load on the elbow joint gradually decreased while the angle of the shoulder joint increased. According to the analysis of the motion and torque data of the joints of the participant, the following can be con- –10 cluded: (1) the kinematic angle variations of joints can be used as the basis for the calculation of the extension value –20 01 2 3 4 4.5 of the actuator of the exoskeleton assistance system; (2) the Time (s) rotational speed of joints represents the basic movement speed of the actuator; (3) the motion and torque data of joints Flexion can be used as the basis for the selection of the energy power Side up Rotation of the exoskeleton assistance system; and (4) the torque and motion of the joints of the upper and lower limbs can be used Figure 7: Torque variations of the three DOFs of the shoulder joint. as the basis for the design of the exoskeleton structure and Angle (°) Torque (Nm) Liing height (mm) Torque (Nm) Applied Bionics and Biomechanics 5 assistance mode. The next section describes the design of the exoskeleton system based on the preceding analysis. 4. Exoskeleton System Design According to the kinematics and analysis of the joint torques in 5 13 specific scenarios presented in the previous section, the exoskel- 6 H eton assistance mode, structure of the limbs and joints, and corresponding electrical system are designed in this section. 4.1. Exoskeleton Assisting Mode. According to the torque data of each joint obtained in Section 3, the torque amplitude of each joint in the lower limb is higher than that of each joint in the upper limb. Moreover, there exist significant differ- ences in the muscle groups that drive each joint [29, 30], and the volume of lower limbs is much larger than that of the upper limbs. Considering that the ankle joint bears a Figure 9: Exoskeleton structure design. 1, upper arm bandage. 2, small torque when lifting a load, this paper focuses only on Bowden line connector of the elbow joint. 3, Bowden line wire the analysis of the shoulder, elbow, hip, and knee joints. rope of the elbow joint. 4, Bowden line anchor of the elbow joint. The average muscle volume that drives each joint of the 5, forearm bandage. 6, Bowden line sheath of the knee joint. 7, lower limb is significantly higher than that which drives each Bowden line wire rope of the knee joint. 8, Bowden line anchor of joint in the upper limb. Additionally, due to the gravity of the the knee joint. 9, Bowden line sheath of the elbow joint. 10, motor upper body, the flexion DOF of the joints in the lower limbs system of the elbow joint. 11, hydraulic cylinder. 12, hydraulic was found to require relatively less effort than the extension pipe. 13, hydraulic energy system. 14, waist fixation block. 15, DOF when the participant tried to squat, so the lower limbs Bowden line wire rope of the hip joint. 16, Bowden line anchor of of the exoskeleton can be designed as passive spring the hip joint. 17, exoskeleton frame. 18, spring for the knee joint. assistance equipment. Consequently, the lower limbs must 19, pulley. 20, spring for the hip joint. overcome the tension of a certain spring, which can also be used as the preload when lifting the load. In contrast, the upper limbs must provide active assistance due to the smaller muscle volume and heavy load imposed on the upper limbs. avoids the complicated DOF fitting for the shoulder joint, From the perspective of the assistance mode, the upper and the energy system is set outside the body. The specific limb has 7 DOFs. Previous studies [31, 32] usually tried to process of the lifting action is presented in Figure 10. set the number of DOFs in the upper limb as high as possible, It can be seen from Figures 10(a) and 10(b) that when which would not only increase the mass of the exoskeleton the wearer squats and starts to lift the load, the joints of but also introduce difficulty to posture detection and control; the lower limbs are in a flexion state, and the hip and knee moreover, this makes it difficult to guarantee the effectiveness joints must overcome the tension of the spring, which is of the assistance of the exoskeleton. Therefore, in this study, fixed at the back of the wearer. Additionally, the shoulder the assistance mode was designed for the specific scenario of joint is also in a flexion state to grab the load, and the lifting a load in situ. Based on the data analysis conducted in hydraulic cylinder then needs to be extended to support the previous section, the forces on the shoulder and elbow the upper arm. As shown in Figures 10(b) and 10(c), the joints are mainly focused on the flexion DOF; therefore, the spring preload acting on the hip and knee joints transforms assistance of the exoskeleton should also be emphasized for the power to joint extension and assists the wearer to com- the flexion DOF. Additionally, due to the lifting of the load plete the upright action. Figures 10(c) and 10(d) reveal that in situ, the driving energy system of the exoskeleton can be the lower limbs remain upright, and the elbow and shoulder fixed outside the wearer’s body, which could reduce the joints begin to flex with assistance from the motor-driven burden caused by the weight of the energy system. According Bowden line and the hydraulic cylinder-driven hydraulic to the preceding analysis, the design of the exoskeleton is energy system, respectively, thereby completing the entire illustrated in Figure 9. lifting movement process. As can be seen from Figure 9, the lower limb of the The following sections provide the details of the design exoskeleton was designed as passive spring assistance. The and parameter calculation based on the assistance mode spring is connected to the hip and knee joints by Bowden and motion data of each joint. lines, which can flexibly transfer the force. Additionally, the springs extend when the wearer squats to grab the load, 4.2. Design of Elbow Joint Assistance. Based on the physical which provides the preload for the extension of the hip and measurements of the participants, the length of the upper knee. The elbow joint is assisted by a motor-driven Bowden arm was set as Ls = 350 mm and the length of the forearm line, and the shoulder joint is assisted by a hydraulic cylinder was set as Lq = 400 mm. The other parameters were set that pushes the upper arm. There are 2 DOFs with the upper according to the anchor position of the Bowden line on the limbs, and the advantage of this structure mode is that it 6 Applied Bionics and Biomechanics (a) (b) (c) (d) Figure 10: The process of lifting a load: (a) initial state; (b) squatting state; (c) lifting load to the upright position; (d) lifting the load to the target position. upper limb, namely, L1 = 100 mm, L2 = 200 mm, L3 = 150 after 3.8 s), which consisted of elbow flexion (2.1~3.2 s) and mm, L4 = 150 mm, and L5 = 80 mm. shoulder flexion (3.2~3.8 s). Moreover, the use of simulta- The motor parameters were obtained based on the struc- neous elbow and shoulder flexion is simulated in Figure 13. ture mode and force analysis. According to Figure 6, the ′ We set course (A) as from position (c)~(c )~(d), and course angle and torque of the elbow joint changed little in the first (B) as from position (c)~(d), which are shown as Figure 13. lifting phase; however, in the second lifting phase, the angle ° ° of the elbow joint increased from 10 to 55 , and the torque 0:5· G · Lq · sin α + β · r ðÞ power of the elbow joint also increased from 10 to 18 Nm. M = pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi : ð5Þ 2 2 cos ω · L4 +L5 · η Thus, the maximum torque of the elbow joint was calculated according to the second lifting phase. The calculation of elbow joint assistance is based on Figure 11 (in which all Figure 13 exhibits the calculation results of the output the symbols are defined), as follows: torque with different flexion courses. In course (A), the max- imum torque was found to be 33 Nm when the angle of elbow ω = arctg L5∕ L4 : ð1Þ ðÞ flexion increased to 55 , which was larger than that of course (B). However, the maximum torque required by the motor- According to Equation (1) and L3 = L4, the following can reducer was considered as 33 Nm. Additionally, the maxi- be calculated: mum rotational speed of the elbow joint was found to be 40.9 /s, as calculated in Section 3.2. The linear velocity of ω = β/2 + ω : ð2Þ the wire rope in the Bowden line should therefore be greater than 86.4 mm/s, so its rotational speed should be no less than Then, the assisting torque M acting on the elbow joint 0.28 r/s. According to the torque and speed calculations, a by Bowden line is calculated as follows: MAXON servo motor RE65 was selected as the motor for pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi elbow joint assistance, a GP81A model was selected as the 2 2 M = F · cos ω · L4 +L5 : ð3Þ z z reducer, and a Decathlon wire (diameter: 1.5 mm) was selected as the Bowden wire. The load torque M acting on elbow joint is defined as follows: 4.3. Design of Shoulder Joint Assistance. The shoulder joint is promoted by hydraulic cylinders. The pressure of the M = G · Lq · sin α + β : ð4Þ hydraulic energy system was selected as 16 MPa. Therefore, ðÞ the size and stroke of the hydraulic cylinder were calculated The tension force F of the Bowden line while in static or and selected as follows based on the kinematics and dynam- in slow motion can be calculated based on Equation (3) and ics of the upper limb. According to the initial state of the Equation (4). hydraulic cylinder shown in Figure 11, the distance between the two fulcrums is calculated as follows: In this study, the load was set as 20 kg, and the transmis- sion efficiency η of the motor–reducer–rotor–Bowden line pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi system shown in Figure 12 was set as 60%. The radius r of 2 2 2 2 Lyd = L1 +L2 = 100 + 200 =223:6mðÞ m : ð6Þ the rotor is 50 mm. Therefore, the output torque M at the end of the reducer for one arm is given by Equation (5), and the variation process is presented in Figure 13. When the wearer squats and starts to lift the load, the nd According to Figure 6, the time of the 2 lifting phase hydraulic cylinder has the largest elongation, and the dis- was from 2.1 s to 3.8 s (the participant was in a static state tance between the two fulcrums is as follows: Lyd Applied Bionics and Biomechanics 7 Shoulder Waist joint Upper arm Ls Upper Bowden fulcrum Ls-200+L1 line sheath Ls L2 L3 Hydraulic cylinder Bowden line wire rope Lower fulcrum L1 L4 Anchor Elbow joint z2 point L5 Forearm z1 Lq (a) (b) L7 z1 F Fy L7 𝛼 F F z2 50° 𝜔 75° Lp L8 Fx z1 𝛾 𝜔 55° F Load z2 𝜔 Ly 10° Load (c) (d) Figure 11: Force analysis of upper limb movement: (a) initial state; (b) squatting state; (c) lifting load to the upright position; (d) lifting load to the target position state. qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 2 2 Lyc = Ls − L2 + L1 +Ls = 250 + 350 =430:1mm : ðÞ ðÞ Reducer ð7Þ Motor Therefore, it can be concluded that the extension of the hydraulic cylinder should be at least 206.5 mm. The distance from the lower fulcrum of the hydraulic cyl- inder to the shoulder joint Lp is as follows: Figure 12: Schematic illustration of Bowden line driven by motor. pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 2 2 Lp = L1 +Ls = 100 + 350 =364 mm : ð8Þ ðÞ According to the second lifting phase, some parameters are defined as follows: Lyc Rotor Bowden line Wire rope Shoulder flexion 8 Applied Bionics and Biomechanics 34 0:5∙G ·L7 (cʹ) F = : ð12Þ ζ∙L8 · sinðÞ δ −ðÞ π∕ 2 − α 30 If the pressure P of the hydraulic system is 16 MPa, the calculation of the inner diameter d of the hydraulic cylinder is as follows: Elbow and shoulder rffiffiffiffiffiffiffiffiffi flexion simultaneously (d) d =2 : ð13Þ P · π Two different lifting courses were also calculated. The maximum force of the hydraulic cylinder occurred during (c) elbow flexion in course (A), and its maximum value was larger than that of course (B). After calculation and compar- 2.1 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8 ison based on Figure 15, the diameter of the piston of the Time (s) hydraulic cylinder should not be less than 10.9 mm, and Figure 13: Torque variation at the end of the retarder with different 12 mm was selected for the design. Additionally, the maxi- flexion courses. Course (A) represents sequential elbow flexion and mum rotational speed of the shoulder joint was found to be shoulder flexion. Course (B) represents simultaneous elbow flexion 41.7 /s, as calculated in Section 3.2. The linear velocity of and shoulder flexion. the hydraulic cylinder should therefore be greater than 106.7 mm/s. Then, based on the extension speed of the hydraulic cylinder and the diameter of its piston, the flow (i) The arm of load L7 and the arm of assisting force L8 should not be less than 0.72 L/min. Consequently, the design are as follows: of the hydraulic system can be obtained according to the calculation. L7 = Ls · sin α + Lq · sin α + β , ðÞ 4.4. Design of Lower Limb Assistance. According to the struc- ð9Þ ture mode presented in Section 4.1, the passive assistance L8 =ðÞ Ls − L2 · sin α mode was adopted for the hip and knee joints, for which the spring and wire rope in series were primarily used for assistance. As shown in Figure 16, one side of each wire rope was fixed on the hip and knee joints, respectively, and the (ii) The length Ly of the hydraulic cylinder in the second other side was connected to the springs by pulleys. According lifting phase is as follows: to Figure 8, the maximum overcoming torques of the hip and knee joints in the first assisting phase are close to 95 Nm and 40 Nm, respectively. The parameters of the springs selected qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi for the hip and knee joints are reported in Table 1, and the Ly = ðÞ Ls − L2 +Lp −2LðÞ s − L2 · Lp · cosðÞ α + γ ð10Þ springs could, respectively, provide 796 N and 280 N of force for the extension of the joints. In addition, the maximum forces transferred by the Bowden lines to the hip and knee joints are, respectively, 47.76 Nm and 19.6 Nm, leading to a passive assistance efficiency of close to 50%. (iii) The angle δ between the upper arm and the upper fulcrum of the hydraulic cylinder is as follows: 4.5. Design of Electrical System. The electrical system of the exoskeleton includes a sensor unit, core processing unit, and execution unit. Among them, the sensor unit is mainly 2 2 ðÞ Ls − L2 +Ly − Lp composed of internal measurement units (IMUs) and δ = arccos ð11Þ 2 · Ly · Ls − L2 ðÞ encoders. The IMUs are arranged along the sagittal plane of the thigh and calf. The encoders are placed at the hip joint ° and knee joint. The IMUs and encoders are primarily used According to Figure 14, the angle δ declines from 89.8 to ° to judge the posture of the lower limbs, which provides the 67 in the second lifting phase, which is the basis for the control criterion for upper limb assistance. An NI sbRIO- calculation of the force of the hydraulic cylinder. 9651 core processing unit was adopted as the bottom unit for data acquisition and information processing, and Lab- (iv) Hydraulic cylinder output: VIEW software was selected as the development environ- ment of the upper computer. The execution unit was The working efficiency ζ of the hydraulic system was set as 50%, and the output force F of the hydraulic cylinder for mainly divided into two parts, namely, the motor-driven one arm was calculated according to Figures 11(c) and Bowden line for the assistance of the elbow joint and the 11(d), as follows: hydraulic cylinder-driven upper arm for the assistance of Elbow flexion M (Nm) G Shoulder flexion Applied Bionics and Biomechanics 9 Table 1: Selection of springs for the hip and knee joints. Parameter Knee spring Hip spring Spring material SUS304 SUS304 Spring diameter (mm) 15 15 Spring wire diameter (mm) 3 4 Total laps 20 20 Initial length (mm) 80 100 Last length (mm) 100 120 eton was assembled, as depicted in Figure 18. The prototype can be adapted to wearers with heights of 170-185 cm by adjusting the lengths of the upper and lower limbs. More- 3.2 3.3 3.4 3.5 3.6 3.7 3.8 over, the weight of the exoskeleton is only 6.8 kg, as the Time (s) energy supply of the prototype is fixed outside the body. Figure 14: Variation of angle δ. Additionally, the shoulder-assisting device driven by a hydraulic cylinder is placed on the back of the upper arm; 0.012 thus, there is no interference with the shoulder movements, and the human-exoskeleton interaction is friendlier. Regarding the control of the prototype, as the lower limbs 0.0115 are driven by passive power, the posture control strategy of the upper limbs can also be provided by IMU detection due 0.011 cʹ to the movement relevance of the upper and lower limbs in the first lifting phase. In the second lifting phase, the upper 0.0105 limbs are driven by a motor + Bowden line and a hydraulic cylinder based on the position control. For the control of Elbow and shoulder 0.01 different lifting heights, the control parameters must be flexion together updated according to the actual scenario requirements. 0.0095 5.2. Test Evaluation of Exoskeleton. Atest evaluation of the exoskeleton system was carried out to verify its effectiveness 0.009 2.1 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8 and consisted of both objective evaluation and subjective Time (s) assessment. Regarding objective evaluations, generally, the EMG signals of test participants wearing an exoskeleton Figure 15: Variation of parameter d with different flexion courses. system for single-joint assistance are collected and are then converted into muscle activation information as the evaluation criterion. Instead, aiming at exoskeleton systems for whole- Knee spring Cam Wire rope for body assistance, oxygen consumption data are collected and hip joint Bowden Pulley are characterized by comprehensive and objective evaluation. line sheath Therefore, in the present study, the oxygen consumption of the test participants was adopted as the objective evaluation Wire rope for Hip spring knee joint criterion. Regarding subjective assessments, participants are usually asked to fill out questionnaires; this method was also adopted in the present study. Figure 16: The mode of lower limb assistance. This study was approved by the Ethics Committee of North China University of Technology. The participants the shoulder joint. A MAXON EPOS4 50/15 module was signed an informed consent form to participate in the study. adopted for the motor driver, while MAXON RE65 + GP81 The information about the five participants is reported in A modules were adopted for the motor + reducer. A small Table 2. Regarding the evaluation protocol, five participants hydraulic station (16 MPa + 1 L/min) was adopted for the each lifted a 15 kg load 10 times from the ground to a height hydraulic system. The specific system composition is illus- of 1.5 m both with and without the exoskeleton, and a respi- trated in Figure 17. rometer (COSMED K5) was used to record their oxygen consumption during each lifting of the load. Additionally, 5. Test and Evaluation of Exoskeleton the quiescent oxygen consumption condition of each partic- 5.1. Prototype Wearing of Exoskeleton. Based on the design, ipant was collected. Then, the differences in the oxygen con- calculation, and model selection of the exoskeleton system sumption during the lifting mode and quiescent condition were obtained and were employed as the objective evaluation described in the previous section, a prototype of the exoskel- Elbow flexion (°) d (m) Upperarm adjustment Forearm adjustment 10 Applied Bionics and Biomechanics Motor Driver IMU Hydraulic station Encoder NI sbRIO-9651 Upper computer Figure 17: Schematic diagram of the exoskeleton electrical system. Upper arm Bowden line anchor bracket Hydraulic cylinder Forearm bracket Strap Drive system Encoder of Bowden line Lower limb bracket IMU (a) igh adjustment (b) (c) Figure 18: Exoskeleton prototype: (a) profile of exoskeleton prototype; (b) adjustment of upper limbs; (c) adjustment of lower limbs. data. For subjective assessment, each participant filled out a intention of use (IU, 0~10), perceived ease of use (PEU, questionnaire at the end of the test to provide their subjective 0~10), and facilitating condition (FC, 0~10). Table 3 shows impression of the exoskeleton from five aspects, namely, the mean values and minimum values of the five participants’ perceived usefulness (PU, 0~10), side effect (SE, -10~0), questionnaires. Applied Bionics and Biomechanics 11 (1) The motion and torque of joints were analysed based Table 2: The information about the five participants. on data collected from a 3-D motion capture system Participant Age Height (mm) Weight (kg) and dynamometer treadmill system. The rotational P1 32 175 65 scope, speed, and torque of the joints when lifting a load in the sagittal plane were obtained and used as P2 36 185 80 the basis for the system design of an exoskeleton P3 28 182 78 P4 24 177 75 (2) Based on the motion of the joints and a muscle layout P5 30 176 75 analysis, an exoskeleton for use when lifting a load in situ was designed as including active assistance for upper limbs and passive assistance for lower limbs. Table 3: Subject scores of the exoskeleton. Regarding the assistance of upper limbs, the energy system was designed to be placed outside the body, PU SE IU PEU FC Total and the assistance mode of the hydraulic cylinder Mean 7.2 -3.1 8.1 8.3 6.5 27 avoids interference between the exoskeleton and the Minimum 5.5 -5 7.5 7 5.5 20.5 body (3) Both objective and subjective methods were adopted for the evaluation of the designed exoskeleton. The median value of oxygen consumed by lifting a load ten times with the assistance of the exoskeleton was found to be reduced by 9.45% as compared with that in the absence of the exoskeleton. Additionally, the subjective feelings of test participants regarding the 20 exoskeleton also proved its effectiveness In future research, more details regarding the use scenarios of the exoskeleton will be included to design a structure and P1 P2 P3 P4 P5 Total energy system, and the assistance efficiency will be promoted via the optimization of human-exoskeleton interaction. Without exoskeleton With exoskeleton Data Availability Figure 19: Oxygen consumption of five participants. The data used to support the findings of this study are avail- The oxygen consumption of the five participants with able from the corresponding author upon request. and without the exoskeleton is presented in Figure 19. The median value of oxygen consumption was found to decrease Conflicts of Interest by 11.3%, 6.79%, 6.28%, 10.4%, 11.26%, and 13.86% of five participants wearing exoskeletons, respectively. Regarding The authors declare that they have no conflicts of interest. the values for the entire experiment, the median value of total oxygen consumption of five participants with exoskeletons Acknowledgments reduced by 9.45% as compared with the absence of the exoskeleton, which demonstrates its assistance effectiveness. This study was supported by the National Natural Science Moreover, as reported in Table 3, the five participants Foundation of China (Grant No. 51775004). provided subjective evaluations of the exoskeleton from the aspects of PU, SE, IU, PEU, and FC, and the results indicate References that the participants found the system helpful for decreasing physical fatigue. While no participant expressed any fear or [1] C. Qu, B. Wu, H. J. Chen, C. C. Yu, and F. Shen, “Upper limb exoskeleton mirror image rehabilitation robot system based on concern about wearing the exoskeleton, the score of FC was somatosensory control,” Journal of Mechanical Engineering, lower than the scores of other aspects because the prototype vol. 29, no. 20, pp. 2484–2489, 2018. is slightly difficult to wear and take off. Consequently, both [2] J. T. Yao, H. L. Li, K. B. Cao, X. Chen, P. Zhou, and Y. Zhao, objective evaluation and subjective assessment verified the “Design and analysis of flexible wearable wrist power glove,” effectiveness of the exoskeleton. Journal of Mechanical Engineering, vol. 54, no. 19, pp. 1–9, 6. Conclusions [3] D. Wang, K. Gui, H. S. Cao, and D. G. Zhang, “Auxiliary robot exoskeleton based on EMG control,” Journal of Yanshan Uni- This paper presented the complete methodology, design, and versity, vol. 42, no. 3, pp. 219–224, 2018. test evaluation of an exoskeleton for use in the scenario of [4] P. N. Kooren, J. Lobo-Prat, A. Q. L. Keemink et al., “Design lifting a load in situ. The findings of the investigation can and control of the Active A-Gear: a wearable 5 DOF arm exo- be summarized as follows: skeleton for adults with Duchenne muscular dystrophy,” in Oxygen consumption (ml/kg/min) 12 Applied Bionics and Biomechanics [19] P. T. Chinimilli, Z. Qiao, S. M. Rezayat Sorkhabadi, V. Jhawar, 2016 6th IEEE International Conference on Biomedical Robot- ics and Biomechatronics (BioRob), pp. 637–642, Singapore, I. H. Fong, and W. Zhang, “Automatic virtual impedance 2016. adaptation of a knee exoskeleton for personalized walking assistance,” Robotics and Autonomous Systems, vol. 114, [5] A. Gupta and M. K. O’Malley, “Design of a haptic arm exoskel- pp. 66–76, 2019. eton for training and rehabilitation,” IEEE/ASME Transactions on Mechatronics, vol. 11, no. 3, pp. 280–289, 2006. [20] Z. F. Lerner, G. M. Gasparri, M. O. Bair et al., “An untethered ankle exoskeleton improves walking economy in a pilot study [6] X. Li, Z. Jia, X. Cui, and L. J. Zhang, “Design, analysis and of individuals with cerebral palsy,” IEEE Transactions on Neu- experiment of a non-humanoid arm exoskeleton for lifting ral Systems and Rehabilitation Engineering, vol. 26, no. 10, load,” in 2018 IEEE International Conference of Intelligent pp. 1985–1993, 2018. Robotic and Control Engineering (IRCE), pp. 64–68, Lanzhou, China, 2018. [21] B. K. Dinh, M. Xiloyannis, C. W. Antuvan, L. Cappello, and L. Masia, “Hierarchical cascade controller for assistance mod- [7] D. M. Baechle, E. D. Wetzel, and S. K. Agrawal, “MAXFAS: ulation in a soft wearable arm exoskeleton,” IEEE Robotics and mechatronic arm exoskeleton for firearm aim stabilization,” Automation Letters, vol. 2, no. 3, pp. 1786–1793, 2017. Journal of Mechanisms and Robotics, vol. 8, no. 6, pp. 061013–061020, 2016. [22] W. Wei, Z. Qu, Q. Wang, P. Zhang, and F. Hao, “Design on the Bowden cable-driven upper limb soft exoskeleton,” Applied [8] S. L. Capitani, M. Bianchi, N. Secciani, M. Pagliai, E. Meli, and Bionics and Biomechanics, vol. 2018, Article ID 1925694, 9 A. Ridolfi, “Model-based mechanical design of a passive lower- pages, 2018. limb exoskeleton for assisting workers in shotcrete projection,” Meccanica, vol. 56, no. 1, pp. 195–210, 2021. [23] M. Puchinger, N. B. R. Kurup, T. Keck, J. Zajc, M. F. Russold, and M. Gföhler, “The retrainer light-weight arm exoskeleton: [9] M. Bolignari and M. Fontana, “Design and experimental char- effect of adjustable gravity compensation on muscle activa- acterization of a high performance hydrostatic transmission tions and forces,” in 2018 7th IEEE International Conference for robot actuation,” Meccanica, vol. 55, no. 5, pp. 1169– on Biomedical Robotics and Biomechatronics (Biorob), 1179, 2020. pp. 396–401, Enschede, Netherlands, 2018. [10] E. A. Brackbill, Y. Mao, S. K. Agrawal, M. Annapragada, and [24] D. Choi and J. H. Oh, “Development of the Cartesian arm exo- V. N. Dubey, “Dynamics and control of a 4-DOF wearable cable-driven upper arm exoskeleton,” 2009 IEEE International skeleton system (CAES) using a 3-axis force/torque sensor,” International Journal of Control, Automation and Systems, Conference on Robotics and Automation, 2009, pp. 2300–2305, vol. 11, no. 5, pp. 976–983, 2013. Kobe, Japan, 2009. [25] P. Maurice, S. Ivaldi, J. Babic et al., “Objective and subjective [11] J. C. Perry, J. Rosen, and S. Burns, “Upper-limb powered exo- effects of a passive exoskeleton on overhead work,” IEEE skeleton design,” IEEE/ASME Transactions on Mechatronics, Transactions on Neural Systems and Rehabilitation Engineer- vol. 12, no. 4, pp. 408–417, 2007. ing, vol. 28, no. 1, pp. 152–164, 2020. [12] S. Sridar, Z. Qiao, A. Rascon et al., “Evaluating immediate ben- efits of assisting knee extension with a soft inflatable exosuit,” [26] Z. Li, C. Xu, Q. Wei, C. Shi, and C. Y. Su, “Human-inspired control of dual-arm exoskeleton robots with force and imped- IEEE Transactions on Medical Robotics and Bionics, vol. 2, no. 2, pp. 216–225, 2020. ance adaptation,” IEEE Transactions on Systems, Man, and Cybernetics: Systems, vol. 50, no. 12, pp. 5296–5305, 2018. [13] R. Goljat, J. Babič, T. Petrič, L. Peternel, and J. Morimoto, [27] H. Yu, I. S. Choi, K. L. Han, J. Y. Choi, G. Chung, and J. Suh, “Power-augmentation control approach for arm exoskeleton based on human muscular manipulability,” in 2017 IEEE “Development of a upper-limb exoskeleton robot for refrac- tory construction,” Control Engineering Practice, vol. 72, International Conference on Robotics and Automation (ICRA), pp. 5929–5934, Singapore, 2017. pp. 104–113, 2018. [14] B. Brahmi, M. Saad, M. H. Rahman, and C. Ochoa-Luna, “Car- [28] A. S. Koopman, I. Kingma, G. S. Faber, M. P. de Looze, and tesian trajectory tracking of a 7-DOF exoskeleton robot based J. H. van Dieën, “Effects of a passive exoskeleton on the on human inverse kinematics,” IEEE Transactions on Systems, mechanical loading of the low back in static holding tasks,” Man, and Cybernetics: Systems, vol. 49, no. 3, pp. 600–611, Journal of Biomechanics, vol. 83, pp. 97–103, 2019. [29] K. R. R. Holzbaur, W. M. Murray, G. E. Gold, and S. L. Delp, [15] L. H. Lu, Q. C. Wu, X. Chen, Z. Shao, B. Chen, and H. Wu, “Upper limb muscle volumes in adult subjects,” Journal of Bio- “Development of a sEMG-based torque estimation control mechanics, vol. 40, no. 4, pp. 742–749, 2007. strategy for a soft elbow exoskeleton,” Robotics and Autono- [30] M. Sakamaki, M. G. Bemben, and T. Abe, “Legs and trunk mous Systems, vol. 111, pp. 88–98, 2019. muscle hypertrophy following walk training with restricted [16] K. Y. Wu, Y. Y. Su, Y. L. Yu, C. H. Lin, and C. C. Lan, “A5- leg muscle blood flow,” Journal of Sports Science and Medicine, degrees-of-freedom lightweight elbow-wrist exoskeleton for vol. 10, no. 2, pp. 338–340, 2011. forearm fine-motion rehabilitation,” IEEE/ASME Transactions [31] X. Cui, W. H. Chen, X. Jin, and S. K. Agrawal, “Design of a 7- on Mechatronics, vol. 24, no. 6, pp. 2684–2695, 2019. DOF cable-driven arm exoskeleton (CAREX-7) and a control- [17] M. T. Picchiotti, E. B. Weston, G. G. Knapik, J. S. Dofour, and ler for dexterous motion training or assistance,” IEEE/ASME W. S. Marras, “Impact of two postural assist exoskeletons on Transactions on Mechatronics, vol. 22, no. 1, pp. 161–172, biomechanical loading of the lumbar spine,” Applied Ergo- nomics, vol. 75, pp. 1–7, 2019. [32] S. Głowiński, T. Krzyżyński, S. Pecolt, and I. Maciejewski, [18] K. Gui, U. X. Tan, H. Liu, and D. G. Zhang, “Electromyogra- “Design of motion trajectory of an arm exoskeleton,” Archive phy-driven progressive assist-as-needed control for lower limb of Applied Mechanics, vol. 85, no. 1, pp. 75–87, 2015. exoskeleton,” IEEE Transactions on Medical Robotics and Bionics, vol. 2, no. 1, pp. 50–58, 2020. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Applied Bionics and Biomechanics Hindawi Publishing Corporation

Method, Design, and Evaluation of an Exoskeleton for Lifting a Load In Situ

Loading next page...
 
/lp/hindawi-publishing-corporation/method-design-and-evaluation-of-an-exoskeleton-for-lifting-a-load-in-ADBxbTU4oe
Publisher
Hindawi Publishing Corporation
Copyright
Copyright © 2021 Xin Li 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
10.1155/2021/5513013
Publisher site
See Article on Publisher Site

Abstract

Hindawi Applied Bionics and Biomechanics Volume 2021, Article ID 5513013, 12 pages https://doi.org/10.1155/2021/5513013 Research Article Method, Design, and Evaluation of an Exoskeleton for Lifting a Load In Situ Xin Li , Weihao Li , and Qiang Li School of Mechanical and Materials Engineering, North China University of Technology, Beijing 100144, China Correspondence should be addressed to Xin Li; lixin2020@ncut.edu.cn Received 15 February 2021; Revised 22 April 2021; Accepted 7 May 2021; Published 25 May 2021 Academic Editor: Andrea Cereatti Copyright © 2021 Xin Li 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. Due to the unclear application scenarios and force analysis of exoskeletons, there exists a research gap in exoskeleton design. This paper presents a design method and realization of an exoskeleton for a specific scenario of lifting a load in situ. Firstly, the lifting motion process and its data were collected based on a 3-D motion capture system and dynamometer treadmill system. Then, the variations of the torque and motion of each joint were obtained from the data analysis, based on which an active assistance mode for upper limbs and a passive assistance mode for lower limbs were demonstrated. In this design, the hydraulic cylinder for shoulder assistance, the motor for elbow assistance, and the spring for lower limb assistance were calculated and selected according to the motion and torque of each joint. Finally, subjective and objective methods were used to evaluate the exoskeleton based on the results of five test participants, and the median oxygen consumption of the whole test by lifting a load ten times with the assistance was found to be reduced by 9.45% as compared with that in the absence of the exoskeleton. 1. Introduction conduct lifting. Generally, active assistance and passive assistance [8] are two common driving modes, and active The rapid development of internet technology has not only assistance modes can be further divided into hydraulic [9], introduced convenience to daily life but has also promoted motor-driven [10, 11], and pneumatic [12, 13] modes. The the expansion of the logistics industry. Robots have been power systems of hydraulic or pneumatic driving modes intro- applied in some logistics tasks, such as handling, grabbing, duce additional load to the body due to their large volume; in and placing items with regular shapes; however, robots can- contrast, a motor can be placed on the back of the body. not replace human beings for the handling of goods with Among the limb assistance components, assistance for upper limbs (shoulder [14], elbow [15], and wrist [16] joints), the uncertain shapes, sizes, or weights. There are nearly 50 mil- lion logistics employees in China, and many (<40 years old) waist joint [17], and lower limbs (hip [18], knee [19], and who work under high-intensity conditions without limb ankle [20] joints) has been developed. Moreover, some protection have suffered from tenosynovitis, lumbar disc exoskeletons, called exosuit, have no rigid frame [21]. An exo- herniation, or other diseases. Therefore, it is necessary to pro- suit is driven by a motor and a Bowden cable that is fixed on vide protection to ensure the health of the limbs and joints of the end of the motor, and force can be transferred to any limb these workers. joints by the Bowden cable [22]. From the perspective of con- An exoskeleton is a functional device that attaches to the trol methods, electromyography (EMG) signals [23] or end human body to assist specific joints to protect or strengthen force/torque detection [24] are usually regarded as the control the body. According to the different application scenarios of input. Regarding exoskeleton evaluation, the subjective feeling products, the modes of exoskeletons can be generally of wearers and the results of an objective test have been illus- categorized into rehabilitation [1–5], industry [6], and military trated to evaluate a passive exoskeleton [25]. [7] applications; for industry and military applications in par- As is evident from the preceding review, there are numer- ticular, the exoskeleton supplements normal human power to ous structural, driving, and control modes that can be adopted 2 Applied Bionics and Biomechanics in exoskeleton design. However, exoskeletons are not general- motion, and the assistance performance must be included as ized devices; the ignorance of the influence of the working design inputs. The design and analysis process is illustrated in scenario or working load on human limbs will lead to the Figure 1. inapplicability of exoskeleton design. For example, rehabilita- According to the figure, the design process of the exoskel- tion exoskeletons are usually oriented to patients with physical eton can be divided into the following five steps: the input of disabilities, who rely on the exoskeleton to provide fixed the working condition, motion capture, data analysis, system movements to perform rehabilitation and physical therapy design, and test evaluation. First, the input of the working on the limbs. Therefore, the emphasis of rehabilitation exo- condition requires the analysis of exoskeleton application skeletons is usually placed on the movement of specificjoints. scenarios, such as the working environment, load profile, The portability of wear and weight are two aspects of and human characteristics, based on which misunderstand- exoskeleton product. In contrast, for industrial or logistics ings of exoskeleton design can be avoided. Second, the torque applications, the target is normal people with a certain labor and motion of the joints should be tested based on 3-D intensity. In this case, comfortable wear is almost as important motion capture equipment, from which the values of the as the effectiveness of assistance; otherwise, unfriendly human- torques, angles, and muscle activity of human limbs can be exoskeleton interaction would directly affect the wearer’swork obtained. Third, the joint torque data can guide the maxi- efficiency. Moreover, the design of exoskeletons requires mum capacity and selection of joints for exoskeleton design, analysis to determine which joints or limbs require assistance and the data of joint motion can be used to determine the based on the application scenario, which has been omitted in speed and motion range of limbs. Fourth, the energy and traditional research. drive systems are designed based on the joint torque data There has been less research on exoskeletons for use in and assistance efficiency. Additionally, the structure and industry or logistics applications than on exoskeletons for mechanical system are designed based on the joint motion use in rehabilitation. Li et al. [26] presented an active dual- data. Then, the exoskeleton prototype system is assembled arm exoskeleton that can be adapted to various environments based on the structure and electrical system. Finally, the by adjusting the force and impedance adaptation, such as results of objective tests and subjective feelings about the lifting loads or rehabilitation training. Yu et al. [27] illus- exoskeleton prototype can be evaluated. In particular, the trated an upper-limb exoskeleton for refractory construction, assistance efficiency of the prototype should be compared which can be used to reduce the physical fatigue of operators to the design values, based on which the system design can resulting from long hours of working with heavy loads; how- be optimized. ever, this active exoskeleton has the deficiencies of a greater self-weight and unfitness for walking. Koopman et al. [28] 3. Motion Capture and Data Analysis of Lifting presented a light and convenient exoskeleton for lumbar Load In Situ protection in logistics work. Dinh et al. [21] illustrated an exosuit that reduces the muscular effort required to lift 1 kg 3.1. Experimental Scenario. The scenario considered in this by 48.3%. Picchiotti et al. [17] compared two commercially research was a logistics sorting operator lifting a load from available postural assist exoskeletons and reported that there the ground to a certain height in a fixed operation area. The is no significant biomechanical benefit regarding the joint load mass was set as 20 kg, and the lifting heights were, flexion angles and moment arms for lifting a given load. As respectively, set as 1 m and 1.5 m. To obtain clear and com- is evident, although passive exoskeletons or exosuits are plete data on the lifting process, a 3-D motion capture system characterized by improved human-machine interaction, they (Phasespace, Impulse X2E) was adopted to collect data on the are not suitable for large loads. Consequently, it is important lifting movement of the whole body and each joint; 36 to consider the application scenario and human-machine marker points were stuck on the body of the wearer, and 10 interaction as the design bases. cameras captured the motion. Additionally, a dynamometer This study presents a complete method of exoskeleton treadmill system (Bertec FIT, FITITC-11-20L) was used to design that is oriented to packing and lifting in logistics work. obtain the plantar force (see Figures 2 and 3). Then, Open- The remainder of this research is organized as follows. Section Sim (V4.0) software was used to collect data on the motion 2 presents the design and analysis process of the exoskeleton and plantar force, based on which the motion of the limbs under the suggested scenario. Section 3 illustrates the motion and the torque of the joints of the test participants were capture and data analysis regarding the movement of entire obtained. limbs, based on which the kinematics and torque of joints are obtained. Section 4 reports the calculation and design of 3.2. Analysis of Experimental Data. Figure 3 depicts the the exoskeleton, which consists of elbow, shoulder, and lower experimental scenario of the lifting test. Data were collected limb assistance, as well as an electrical system. Section 5 illus- from the 3-D motion capture system and dynamometer trates the prototype and experiments, the results of which treadmill system to determine the motion of the limbs and verify the effectiveness of the design process. the torque of the joints of the participant, after which the data were imported into OpenSim software to restore the partici- pant’s motion and determine the motion variation process, as 2. Exoskeleton Design and Analysis Process shown in Figure 4. This process was primarily conducted to As mentioned in the previous section, exoskeletons are not uni- verify the effectiveness of the data collection of the 3-D versal products; thus, the working environment, the wearer’s motion capture system and to provide support for the inverse Applied Bionics and Biomechanics 3 Maximum Energy Subjective Joint capacity Environment and evaluation torque Selected drive joints Assisting Working system Assisting Oxygen Human Motion Exoskeleten efficiency condition effect consumption Structure Speed Joint and Load motion EMG data mechanical Range system Input of working condition Motion capture Data analysis System design Test and evaluation Figure 1: The design and analysis process of an exoskeleton. kinematics solution of the torques of the joints in the subse- quent step. Figure 5 reveals the height variation of the lifted load collected by the 3-D motion capture system, which meets Data acquisition the design requirements. Figure 6 presents the variations of 3-D motion capture camera and timing the angles of the hip, knee, ankle, shoulder, and elbow joints system in the extension/flexion degree of freedom (DOF) when lift- Marker point ing the load. The first lifting phase was from the ground to a height of 1 m, and the second lifting phase was from the Load height of 1 m to the height of 1.5 m. As shown in Figure 6, Dyanmometer Computer during the first lifting phase, the lower limbs of the partici- treadmill pant gradually changed from the bending state to the upright ° ° state; specifically, the hip joint changed from 90 to -10 , the Figure 2: Schematic diagram of the lifting test. ° ° knee joint changed from -130 to 0 , and the ankle joint chan- ° ° ged from 32 to -7 . During the second lifting phase, there was almost no change in any joint of the lower limbs. Regarding the upper limbs, the angle of the shoulder joint was 90 when the participant grabbed the load on the ground; it was reduced to 50 after the first lifting phase, and it finally increased to 75 during the second lifting phase. The angle 3-D motion capture camera of the elbow joint changed little during the first lifting phase ° ° and increased from 10 to 55 during the second lifting phase. Additionally, the rotational speeds of the limb joints were Marker point obtained based on Figure 6. In the first lifting phase, the average speeds of the hip, knee, and ankle joints were found ° ° ° to be 75 /s, 76.47 /s, and 18.82 /s, respectively. Regarding the Dynamometer upper limbs, the rotational speeds of the elbow and shoulder treadmill joints in the second lifting phase presented faster variations than in the first lifting phase, and the maximum rotational ° ° speeds were about 40.9 /s and 41.7 /s, respectively. These kine- Load matic values of the limbs and joints provide the basis of the structural and system design of the exoskeleton system. As lifting in situ is primarily completed in the sagittal plane, the extension/flexion DOF of each joint provides more Figure 3: Experiment of lifting test. support for lifting. As shown in Figure 7, the torque value of the flexion DOF was much higher than that of the side-up and rotation DOFs. Therefore, this paper focuses on the analysis of the extension/flexion DOF of each joint. Figure 8 presents the torque variations of the hip, knee, ankle, shoulder, and elbow joints in the extension/flexion DOF during the lifting process. The torque amplitudes of the joints in the lower limb were found to be higher than those of the joints in the upper limb. In addition, the torque Figure 4: The motion animation of the participant restored by values of the upper-limb joints are positive because the OpenSim software. rotational direction of torque acting on the elbow and shoul- der is clockwise; specifically, the latissimus dorsi behind the Height (mm) 4 Applied Bionics and Biomechanics 1600 50 1200 nd 2 liing height st 1 liing height 600 –50 01 2 3 4 4.5 –100 0 1 2 3 4 4.5 Time (s) Time (s) Figure 5: The height variation of the lifted load. Hip Shoulder Knee Elbow Ankle Figure 8: Torque variations of each joint during lifting. shoulder is mainly responsible for its flexion, and the bicipital muscle on the upper limb is mainly responsible for elbow flexion. Regarding the lower limbs, the direction of DOFs of –50 the hip and ankle joints is extension during the entire lifting process, and the direction of torque driven by the corre- sponding muscle to the joint is counterclockwise; thus, the –100 st Holding 1 liing values of their torques are negative. Regarding the knee, phase nd phase 2 liing phase torque assists the extension DOF of the joint in the first lift- –150 ing phase, the direction of which is clockwise, so the value 01 2 3 4 4.5 is positive. Subsequently, the value changes to a negative Time (s) value because the direction of DOF changes to flexion, which Hip Shoulder changes the direction of the joint torque to counterclockwise. Knee Elbow Regarding the torque amplitude, the value of the hip joint Ankle ranged from -95 to -70 Nm, that of the knee joint changed from 40 to -70 Nm, and that of the ankle joint ranged from Figure 6: Variations of joints’ angles during lifting. -45 to -10 Nm throughout the entire lifting phase. During the first lifting phase, the torque of the shoulder joint was found to increase from 17 to 30 Nm, and that of the elbow joint increased from 0 to 10 Nm. During the second lifting phase, the torque of the shoulder joint was found to continu- ally increase from 30 Nm to nearly 50 Nm, while the torque of the elbow joint increased from 10 to 18 Nm and then decreased gradually after reaching the peak; this occurred because the load was closer to the participant, and the load on the elbow joint gradually decreased while the angle of the shoulder joint increased. According to the analysis of the motion and torque data of the joints of the participant, the following can be con- –10 cluded: (1) the kinematic angle variations of joints can be used as the basis for the calculation of the extension value –20 01 2 3 4 4.5 of the actuator of the exoskeleton assistance system; (2) the Time (s) rotational speed of joints represents the basic movement speed of the actuator; (3) the motion and torque data of joints Flexion can be used as the basis for the selection of the energy power Side up Rotation of the exoskeleton assistance system; and (4) the torque and motion of the joints of the upper and lower limbs can be used Figure 7: Torque variations of the three DOFs of the shoulder joint. as the basis for the design of the exoskeleton structure and Angle (°) Torque (Nm) Liing height (mm) Torque (Nm) Applied Bionics and Biomechanics 5 assistance mode. The next section describes the design of the exoskeleton system based on the preceding analysis. 4. Exoskeleton System Design According to the kinematics and analysis of the joint torques in 5 13 specific scenarios presented in the previous section, the exoskel- 6 H eton assistance mode, structure of the limbs and joints, and corresponding electrical system are designed in this section. 4.1. Exoskeleton Assisting Mode. According to the torque data of each joint obtained in Section 3, the torque amplitude of each joint in the lower limb is higher than that of each joint in the upper limb. Moreover, there exist significant differ- ences in the muscle groups that drive each joint [29, 30], and the volume of lower limbs is much larger than that of the upper limbs. Considering that the ankle joint bears a Figure 9: Exoskeleton structure design. 1, upper arm bandage. 2, small torque when lifting a load, this paper focuses only on Bowden line connector of the elbow joint. 3, Bowden line wire the analysis of the shoulder, elbow, hip, and knee joints. rope of the elbow joint. 4, Bowden line anchor of the elbow joint. The average muscle volume that drives each joint of the 5, forearm bandage. 6, Bowden line sheath of the knee joint. 7, lower limb is significantly higher than that which drives each Bowden line wire rope of the knee joint. 8, Bowden line anchor of joint in the upper limb. Additionally, due to the gravity of the the knee joint. 9, Bowden line sheath of the elbow joint. 10, motor upper body, the flexion DOF of the joints in the lower limbs system of the elbow joint. 11, hydraulic cylinder. 12, hydraulic was found to require relatively less effort than the extension pipe. 13, hydraulic energy system. 14, waist fixation block. 15, DOF when the participant tried to squat, so the lower limbs Bowden line wire rope of the hip joint. 16, Bowden line anchor of of the exoskeleton can be designed as passive spring the hip joint. 17, exoskeleton frame. 18, spring for the knee joint. assistance equipment. Consequently, the lower limbs must 19, pulley. 20, spring for the hip joint. overcome the tension of a certain spring, which can also be used as the preload when lifting the load. In contrast, the upper limbs must provide active assistance due to the smaller muscle volume and heavy load imposed on the upper limbs. avoids the complicated DOF fitting for the shoulder joint, From the perspective of the assistance mode, the upper and the energy system is set outside the body. The specific limb has 7 DOFs. Previous studies [31, 32] usually tried to process of the lifting action is presented in Figure 10. set the number of DOFs in the upper limb as high as possible, It can be seen from Figures 10(a) and 10(b) that when which would not only increase the mass of the exoskeleton the wearer squats and starts to lift the load, the joints of but also introduce difficulty to posture detection and control; the lower limbs are in a flexion state, and the hip and knee moreover, this makes it difficult to guarantee the effectiveness joints must overcome the tension of the spring, which is of the assistance of the exoskeleton. Therefore, in this study, fixed at the back of the wearer. Additionally, the shoulder the assistance mode was designed for the specific scenario of joint is also in a flexion state to grab the load, and the lifting a load in situ. Based on the data analysis conducted in hydraulic cylinder then needs to be extended to support the previous section, the forces on the shoulder and elbow the upper arm. As shown in Figures 10(b) and 10(c), the joints are mainly focused on the flexion DOF; therefore, the spring preload acting on the hip and knee joints transforms assistance of the exoskeleton should also be emphasized for the power to joint extension and assists the wearer to com- the flexion DOF. Additionally, due to the lifting of the load plete the upright action. Figures 10(c) and 10(d) reveal that in situ, the driving energy system of the exoskeleton can be the lower limbs remain upright, and the elbow and shoulder fixed outside the wearer’s body, which could reduce the joints begin to flex with assistance from the motor-driven burden caused by the weight of the energy system. According Bowden line and the hydraulic cylinder-driven hydraulic to the preceding analysis, the design of the exoskeleton is energy system, respectively, thereby completing the entire illustrated in Figure 9. lifting movement process. As can be seen from Figure 9, the lower limb of the The following sections provide the details of the design exoskeleton was designed as passive spring assistance. The and parameter calculation based on the assistance mode spring is connected to the hip and knee joints by Bowden and motion data of each joint. lines, which can flexibly transfer the force. Additionally, the springs extend when the wearer squats to grab the load, 4.2. Design of Elbow Joint Assistance. Based on the physical which provides the preload for the extension of the hip and measurements of the participants, the length of the upper knee. The elbow joint is assisted by a motor-driven Bowden arm was set as Ls = 350 mm and the length of the forearm line, and the shoulder joint is assisted by a hydraulic cylinder was set as Lq = 400 mm. The other parameters were set that pushes the upper arm. There are 2 DOFs with the upper according to the anchor position of the Bowden line on the limbs, and the advantage of this structure mode is that it 6 Applied Bionics and Biomechanics (a) (b) (c) (d) Figure 10: The process of lifting a load: (a) initial state; (b) squatting state; (c) lifting load to the upright position; (d) lifting the load to the target position. upper limb, namely, L1 = 100 mm, L2 = 200 mm, L3 = 150 after 3.8 s), which consisted of elbow flexion (2.1~3.2 s) and mm, L4 = 150 mm, and L5 = 80 mm. shoulder flexion (3.2~3.8 s). Moreover, the use of simulta- The motor parameters were obtained based on the struc- neous elbow and shoulder flexion is simulated in Figure 13. ture mode and force analysis. According to Figure 6, the ′ We set course (A) as from position (c)~(c )~(d), and course angle and torque of the elbow joint changed little in the first (B) as from position (c)~(d), which are shown as Figure 13. lifting phase; however, in the second lifting phase, the angle ° ° of the elbow joint increased from 10 to 55 , and the torque 0:5· G · Lq · sin α + β · r ðÞ power of the elbow joint also increased from 10 to 18 Nm. M = pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi : ð5Þ 2 2 cos ω · L4 +L5 · η Thus, the maximum torque of the elbow joint was calculated according to the second lifting phase. The calculation of elbow joint assistance is based on Figure 11 (in which all Figure 13 exhibits the calculation results of the output the symbols are defined), as follows: torque with different flexion courses. In course (A), the max- imum torque was found to be 33 Nm when the angle of elbow ω = arctg L5∕ L4 : ð1Þ ðÞ flexion increased to 55 , which was larger than that of course (B). However, the maximum torque required by the motor- According to Equation (1) and L3 = L4, the following can reducer was considered as 33 Nm. Additionally, the maxi- be calculated: mum rotational speed of the elbow joint was found to be 40.9 /s, as calculated in Section 3.2. The linear velocity of ω = β/2 + ω : ð2Þ the wire rope in the Bowden line should therefore be greater than 86.4 mm/s, so its rotational speed should be no less than Then, the assisting torque M acting on the elbow joint 0.28 r/s. According to the torque and speed calculations, a by Bowden line is calculated as follows: MAXON servo motor RE65 was selected as the motor for pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi elbow joint assistance, a GP81A model was selected as the 2 2 M = F · cos ω · L4 +L5 : ð3Þ z z reducer, and a Decathlon wire (diameter: 1.5 mm) was selected as the Bowden wire. The load torque M acting on elbow joint is defined as follows: 4.3. Design of Shoulder Joint Assistance. The shoulder joint is promoted by hydraulic cylinders. The pressure of the M = G · Lq · sin α + β : ð4Þ hydraulic energy system was selected as 16 MPa. Therefore, ðÞ the size and stroke of the hydraulic cylinder were calculated The tension force F of the Bowden line while in static or and selected as follows based on the kinematics and dynam- in slow motion can be calculated based on Equation (3) and ics of the upper limb. According to the initial state of the Equation (4). hydraulic cylinder shown in Figure 11, the distance between the two fulcrums is calculated as follows: In this study, the load was set as 20 kg, and the transmis- sion efficiency η of the motor–reducer–rotor–Bowden line pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi system shown in Figure 12 was set as 60%. The radius r of 2 2 2 2 Lyd = L1 +L2 = 100 + 200 =223:6mðÞ m : ð6Þ the rotor is 50 mm. Therefore, the output torque M at the end of the reducer for one arm is given by Equation (5), and the variation process is presented in Figure 13. When the wearer squats and starts to lift the load, the nd According to Figure 6, the time of the 2 lifting phase hydraulic cylinder has the largest elongation, and the dis- was from 2.1 s to 3.8 s (the participant was in a static state tance between the two fulcrums is as follows: Lyd Applied Bionics and Biomechanics 7 Shoulder Waist joint Upper arm Ls Upper Bowden fulcrum Ls-200+L1 line sheath Ls L2 L3 Hydraulic cylinder Bowden line wire rope Lower fulcrum L1 L4 Anchor Elbow joint z2 point L5 Forearm z1 Lq (a) (b) L7 z1 F Fy L7 𝛼 F F z2 50° 𝜔 75° Lp L8 Fx z1 𝛾 𝜔 55° F Load z2 𝜔 Ly 10° Load (c) (d) Figure 11: Force analysis of upper limb movement: (a) initial state; (b) squatting state; (c) lifting load to the upright position; (d) lifting load to the target position state. qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 2 2 Lyc = Ls − L2 + L1 +Ls = 250 + 350 =430:1mm : ðÞ ðÞ Reducer ð7Þ Motor Therefore, it can be concluded that the extension of the hydraulic cylinder should be at least 206.5 mm. The distance from the lower fulcrum of the hydraulic cyl- inder to the shoulder joint Lp is as follows: Figure 12: Schematic illustration of Bowden line driven by motor. pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 2 2 Lp = L1 +Ls = 100 + 350 =364 mm : ð8Þ ðÞ According to the second lifting phase, some parameters are defined as follows: Lyc Rotor Bowden line Wire rope Shoulder flexion 8 Applied Bionics and Biomechanics 34 0:5∙G ·L7 (cʹ) F = : ð12Þ ζ∙L8 · sinðÞ δ −ðÞ π∕ 2 − α 30 If the pressure P of the hydraulic system is 16 MPa, the calculation of the inner diameter d of the hydraulic cylinder is as follows: Elbow and shoulder rffiffiffiffiffiffiffiffiffi flexion simultaneously (d) d =2 : ð13Þ P · π Two different lifting courses were also calculated. The maximum force of the hydraulic cylinder occurred during (c) elbow flexion in course (A), and its maximum value was larger than that of course (B). After calculation and compar- 2.1 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8 ison based on Figure 15, the diameter of the piston of the Time (s) hydraulic cylinder should not be less than 10.9 mm, and Figure 13: Torque variation at the end of the retarder with different 12 mm was selected for the design. Additionally, the maxi- flexion courses. Course (A) represents sequential elbow flexion and mum rotational speed of the shoulder joint was found to be shoulder flexion. Course (B) represents simultaneous elbow flexion 41.7 /s, as calculated in Section 3.2. The linear velocity of and shoulder flexion. the hydraulic cylinder should therefore be greater than 106.7 mm/s. Then, based on the extension speed of the hydraulic cylinder and the diameter of its piston, the flow (i) The arm of load L7 and the arm of assisting force L8 should not be less than 0.72 L/min. Consequently, the design are as follows: of the hydraulic system can be obtained according to the calculation. L7 = Ls · sin α + Lq · sin α + β , ðÞ 4.4. Design of Lower Limb Assistance. According to the struc- ð9Þ ture mode presented in Section 4.1, the passive assistance L8 =ðÞ Ls − L2 · sin α mode was adopted for the hip and knee joints, for which the spring and wire rope in series were primarily used for assistance. As shown in Figure 16, one side of each wire rope was fixed on the hip and knee joints, respectively, and the (ii) The length Ly of the hydraulic cylinder in the second other side was connected to the springs by pulleys. According lifting phase is as follows: to Figure 8, the maximum overcoming torques of the hip and knee joints in the first assisting phase are close to 95 Nm and 40 Nm, respectively. The parameters of the springs selected qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi for the hip and knee joints are reported in Table 1, and the Ly = ðÞ Ls − L2 +Lp −2LðÞ s − L2 · Lp · cosðÞ α + γ ð10Þ springs could, respectively, provide 796 N and 280 N of force for the extension of the joints. In addition, the maximum forces transferred by the Bowden lines to the hip and knee joints are, respectively, 47.76 Nm and 19.6 Nm, leading to a passive assistance efficiency of close to 50%. (iii) The angle δ between the upper arm and the upper fulcrum of the hydraulic cylinder is as follows: 4.5. Design of Electrical System. The electrical system of the exoskeleton includes a sensor unit, core processing unit, and execution unit. Among them, the sensor unit is mainly 2 2 ðÞ Ls − L2 +Ly − Lp composed of internal measurement units (IMUs) and δ = arccos ð11Þ 2 · Ly · Ls − L2 ðÞ encoders. The IMUs are arranged along the sagittal plane of the thigh and calf. The encoders are placed at the hip joint ° and knee joint. The IMUs and encoders are primarily used According to Figure 14, the angle δ declines from 89.8 to ° to judge the posture of the lower limbs, which provides the 67 in the second lifting phase, which is the basis for the control criterion for upper limb assistance. An NI sbRIO- calculation of the force of the hydraulic cylinder. 9651 core processing unit was adopted as the bottom unit for data acquisition and information processing, and Lab- (iv) Hydraulic cylinder output: VIEW software was selected as the development environ- ment of the upper computer. The execution unit was The working efficiency ζ of the hydraulic system was set as 50%, and the output force F of the hydraulic cylinder for mainly divided into two parts, namely, the motor-driven one arm was calculated according to Figures 11(c) and Bowden line for the assistance of the elbow joint and the 11(d), as follows: hydraulic cylinder-driven upper arm for the assistance of Elbow flexion M (Nm) G Shoulder flexion Applied Bionics and Biomechanics 9 Table 1: Selection of springs for the hip and knee joints. Parameter Knee spring Hip spring Spring material SUS304 SUS304 Spring diameter (mm) 15 15 Spring wire diameter (mm) 3 4 Total laps 20 20 Initial length (mm) 80 100 Last length (mm) 100 120 eton was assembled, as depicted in Figure 18. The prototype can be adapted to wearers with heights of 170-185 cm by adjusting the lengths of the upper and lower limbs. More- 3.2 3.3 3.4 3.5 3.6 3.7 3.8 over, the weight of the exoskeleton is only 6.8 kg, as the Time (s) energy supply of the prototype is fixed outside the body. Figure 14: Variation of angle δ. Additionally, the shoulder-assisting device driven by a hydraulic cylinder is placed on the back of the upper arm; 0.012 thus, there is no interference with the shoulder movements, and the human-exoskeleton interaction is friendlier. Regarding the control of the prototype, as the lower limbs 0.0115 are driven by passive power, the posture control strategy of the upper limbs can also be provided by IMU detection due 0.011 cʹ to the movement relevance of the upper and lower limbs in the first lifting phase. In the second lifting phase, the upper 0.0105 limbs are driven by a motor + Bowden line and a hydraulic cylinder based on the position control. For the control of Elbow and shoulder 0.01 different lifting heights, the control parameters must be flexion together updated according to the actual scenario requirements. 0.0095 5.2. Test Evaluation of Exoskeleton. Atest evaluation of the exoskeleton system was carried out to verify its effectiveness 0.009 2.1 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8 and consisted of both objective evaluation and subjective Time (s) assessment. Regarding objective evaluations, generally, the EMG signals of test participants wearing an exoskeleton Figure 15: Variation of parameter d with different flexion courses. system for single-joint assistance are collected and are then converted into muscle activation information as the evaluation criterion. Instead, aiming at exoskeleton systems for whole- Knee spring Cam Wire rope for body assistance, oxygen consumption data are collected and hip joint Bowden Pulley are characterized by comprehensive and objective evaluation. line sheath Therefore, in the present study, the oxygen consumption of the test participants was adopted as the objective evaluation Wire rope for Hip spring knee joint criterion. Regarding subjective assessments, participants are usually asked to fill out questionnaires; this method was also adopted in the present study. Figure 16: The mode of lower limb assistance. This study was approved by the Ethics Committee of North China University of Technology. The participants the shoulder joint. A MAXON EPOS4 50/15 module was signed an informed consent form to participate in the study. adopted for the motor driver, while MAXON RE65 + GP81 The information about the five participants is reported in A modules were adopted for the motor + reducer. A small Table 2. Regarding the evaluation protocol, five participants hydraulic station (16 MPa + 1 L/min) was adopted for the each lifted a 15 kg load 10 times from the ground to a height hydraulic system. The specific system composition is illus- of 1.5 m both with and without the exoskeleton, and a respi- trated in Figure 17. rometer (COSMED K5) was used to record their oxygen consumption during each lifting of the load. Additionally, 5. Test and Evaluation of Exoskeleton the quiescent oxygen consumption condition of each partic- 5.1. Prototype Wearing of Exoskeleton. Based on the design, ipant was collected. Then, the differences in the oxygen con- calculation, and model selection of the exoskeleton system sumption during the lifting mode and quiescent condition were obtained and were employed as the objective evaluation described in the previous section, a prototype of the exoskel- Elbow flexion (°) d (m) Upperarm adjustment Forearm adjustment 10 Applied Bionics and Biomechanics Motor Driver IMU Hydraulic station Encoder NI sbRIO-9651 Upper computer Figure 17: Schematic diagram of the exoskeleton electrical system. Upper arm Bowden line anchor bracket Hydraulic cylinder Forearm bracket Strap Drive system Encoder of Bowden line Lower limb bracket IMU (a) igh adjustment (b) (c) Figure 18: Exoskeleton prototype: (a) profile of exoskeleton prototype; (b) adjustment of upper limbs; (c) adjustment of lower limbs. data. For subjective assessment, each participant filled out a intention of use (IU, 0~10), perceived ease of use (PEU, questionnaire at the end of the test to provide their subjective 0~10), and facilitating condition (FC, 0~10). Table 3 shows impression of the exoskeleton from five aspects, namely, the mean values and minimum values of the five participants’ perceived usefulness (PU, 0~10), side effect (SE, -10~0), questionnaires. Applied Bionics and Biomechanics 11 (1) The motion and torque of joints were analysed based Table 2: The information about the five participants. on data collected from a 3-D motion capture system Participant Age Height (mm) Weight (kg) and dynamometer treadmill system. The rotational P1 32 175 65 scope, speed, and torque of the joints when lifting a load in the sagittal plane were obtained and used as P2 36 185 80 the basis for the system design of an exoskeleton P3 28 182 78 P4 24 177 75 (2) Based on the motion of the joints and a muscle layout P5 30 176 75 analysis, an exoskeleton for use when lifting a load in situ was designed as including active assistance for upper limbs and passive assistance for lower limbs. Table 3: Subject scores of the exoskeleton. Regarding the assistance of upper limbs, the energy system was designed to be placed outside the body, PU SE IU PEU FC Total and the assistance mode of the hydraulic cylinder Mean 7.2 -3.1 8.1 8.3 6.5 27 avoids interference between the exoskeleton and the Minimum 5.5 -5 7.5 7 5.5 20.5 body (3) Both objective and subjective methods were adopted for the evaluation of the designed exoskeleton. The median value of oxygen consumed by lifting a load ten times with the assistance of the exoskeleton was found to be reduced by 9.45% as compared with that in the absence of the exoskeleton. Additionally, the subjective feelings of test participants regarding the 20 exoskeleton also proved its effectiveness In future research, more details regarding the use scenarios of the exoskeleton will be included to design a structure and P1 P2 P3 P4 P5 Total energy system, and the assistance efficiency will be promoted via the optimization of human-exoskeleton interaction. Without exoskeleton With exoskeleton Data Availability Figure 19: Oxygen consumption of five participants. The data used to support the findings of this study are avail- The oxygen consumption of the five participants with able from the corresponding author upon request. and without the exoskeleton is presented in Figure 19. The median value of oxygen consumption was found to decrease Conflicts of Interest by 11.3%, 6.79%, 6.28%, 10.4%, 11.26%, and 13.86% of five participants wearing exoskeletons, respectively. Regarding The authors declare that they have no conflicts of interest. the values for the entire experiment, the median value of total oxygen consumption of five participants with exoskeletons Acknowledgments reduced by 9.45% as compared with the absence of the exoskeleton, which demonstrates its assistance effectiveness. This study was supported by the National Natural Science Moreover, as reported in Table 3, the five participants Foundation of China (Grant No. 51775004). provided subjective evaluations of the exoskeleton from the aspects of PU, SE, IU, PEU, and FC, and the results indicate References that the participants found the system helpful for decreasing physical fatigue. While no participant expressed any fear or [1] C. Qu, B. Wu, H. J. Chen, C. C. Yu, and F. Shen, “Upper limb exoskeleton mirror image rehabilitation robot system based on concern about wearing the exoskeleton, the score of FC was somatosensory control,” Journal of Mechanical Engineering, lower than the scores of other aspects because the prototype vol. 29, no. 20, pp. 2484–2489, 2018. is slightly difficult to wear and take off. Consequently, both [2] J. T. Yao, H. L. Li, K. B. Cao, X. Chen, P. Zhou, and Y. Zhao, objective evaluation and subjective assessment verified the “Design and analysis of flexible wearable wrist power glove,” effectiveness of the exoskeleton. Journal of Mechanical Engineering, vol. 54, no. 19, pp. 1–9, 6. Conclusions [3] D. Wang, K. Gui, H. S. Cao, and D. G. Zhang, “Auxiliary robot exoskeleton based on EMG control,” Journal of Yanshan Uni- This paper presented the complete methodology, design, and versity, vol. 42, no. 3, pp. 219–224, 2018. test evaluation of an exoskeleton for use in the scenario of [4] P. N. Kooren, J. Lobo-Prat, A. Q. L. Keemink et al., “Design lifting a load in situ. The findings of the investigation can and control of the Active A-Gear: a wearable 5 DOF arm exo- be summarized as follows: skeleton for adults with Duchenne muscular dystrophy,” in Oxygen consumption (ml/kg/min) 12 Applied Bionics and Biomechanics [19] P. T. Chinimilli, Z. Qiao, S. M. Rezayat Sorkhabadi, V. Jhawar, 2016 6th IEEE International Conference on Biomedical Robot- ics and Biomechatronics (BioRob), pp. 637–642, Singapore, I. H. Fong, and W. Zhang, “Automatic virtual impedance 2016. adaptation of a knee exoskeleton for personalized walking assistance,” Robotics and Autonomous Systems, vol. 114, [5] A. Gupta and M. K. O’Malley, “Design of a haptic arm exoskel- pp. 66–76, 2019. eton for training and rehabilitation,” IEEE/ASME Transactions on Mechatronics, vol. 11, no. 3, pp. 280–289, 2006. [20] Z. F. Lerner, G. M. Gasparri, M. O. Bair et al., “An untethered ankle exoskeleton improves walking economy in a pilot study [6] X. Li, Z. Jia, X. Cui, and L. J. Zhang, “Design, analysis and of individuals with cerebral palsy,” IEEE Transactions on Neu- experiment of a non-humanoid arm exoskeleton for lifting ral Systems and Rehabilitation Engineering, vol. 26, no. 10, load,” in 2018 IEEE International Conference of Intelligent pp. 1985–1993, 2018. Robotic and Control Engineering (IRCE), pp. 64–68, Lanzhou, China, 2018. [21] B. K. Dinh, M. Xiloyannis, C. W. Antuvan, L. Cappello, and L. Masia, “Hierarchical cascade controller for assistance mod- [7] D. M. Baechle, E. D. Wetzel, and S. K. Agrawal, “MAXFAS: ulation in a soft wearable arm exoskeleton,” IEEE Robotics and mechatronic arm exoskeleton for firearm aim stabilization,” Automation Letters, vol. 2, no. 3, pp. 1786–1793, 2017. Journal of Mechanisms and Robotics, vol. 8, no. 6, pp. 061013–061020, 2016. [22] W. Wei, Z. Qu, Q. Wang, P. Zhang, and F. Hao, “Design on the Bowden cable-driven upper limb soft exoskeleton,” Applied [8] S. L. Capitani, M. Bianchi, N. Secciani, M. Pagliai, E. Meli, and Bionics and Biomechanics, vol. 2018, Article ID 1925694, 9 A. Ridolfi, “Model-based mechanical design of a passive lower- pages, 2018. limb exoskeleton for assisting workers in shotcrete projection,” Meccanica, vol. 56, no. 1, pp. 195–210, 2021. [23] M. Puchinger, N. B. R. Kurup, T. Keck, J. Zajc, M. F. Russold, and M. Gföhler, “The retrainer light-weight arm exoskeleton: [9] M. Bolignari and M. Fontana, “Design and experimental char- effect of adjustable gravity compensation on muscle activa- acterization of a high performance hydrostatic transmission tions and forces,” in 2018 7th IEEE International Conference for robot actuation,” Meccanica, vol. 55, no. 5, pp. 1169– on Biomedical Robotics and Biomechatronics (Biorob), 1179, 2020. pp. 396–401, Enschede, Netherlands, 2018. [10] E. A. Brackbill, Y. Mao, S. K. Agrawal, M. Annapragada, and [24] D. Choi and J. H. Oh, “Development of the Cartesian arm exo- V. N. Dubey, “Dynamics and control of a 4-DOF wearable cable-driven upper arm exoskeleton,” 2009 IEEE International skeleton system (CAES) using a 3-axis force/torque sensor,” International Journal of Control, Automation and Systems, Conference on Robotics and Automation, 2009, pp. 2300–2305, vol. 11, no. 5, pp. 976–983, 2013. Kobe, Japan, 2009. [25] P. Maurice, S. Ivaldi, J. Babic et al., “Objective and subjective [11] J. C. Perry, J. Rosen, and S. Burns, “Upper-limb powered exo- effects of a passive exoskeleton on overhead work,” IEEE skeleton design,” IEEE/ASME Transactions on Mechatronics, Transactions on Neural Systems and Rehabilitation Engineer- vol. 12, no. 4, pp. 408–417, 2007. ing, vol. 28, no. 1, pp. 152–164, 2020. [12] S. Sridar, Z. Qiao, A. Rascon et al., “Evaluating immediate ben- efits of assisting knee extension with a soft inflatable exosuit,” [26] Z. Li, C. Xu, Q. Wei, C. Shi, and C. Y. Su, “Human-inspired control of dual-arm exoskeleton robots with force and imped- IEEE Transactions on Medical Robotics and Bionics, vol. 2, no. 2, pp. 216–225, 2020. ance adaptation,” IEEE Transactions on Systems, Man, and Cybernetics: Systems, vol. 50, no. 12, pp. 5296–5305, 2018. [13] R. Goljat, J. Babič, T. Petrič, L. Peternel, and J. Morimoto, [27] H. Yu, I. S. Choi, K. L. Han, J. Y. Choi, G. Chung, and J. Suh, “Power-augmentation control approach for arm exoskeleton based on human muscular manipulability,” in 2017 IEEE “Development of a upper-limb exoskeleton robot for refrac- tory construction,” Control Engineering Practice, vol. 72, International Conference on Robotics and Automation (ICRA), pp. 5929–5934, Singapore, 2017. pp. 104–113, 2018. [14] B. Brahmi, M. Saad, M. H. Rahman, and C. Ochoa-Luna, “Car- [28] A. S. Koopman, I. Kingma, G. S. Faber, M. P. de Looze, and tesian trajectory tracking of a 7-DOF exoskeleton robot based J. H. van Dieën, “Effects of a passive exoskeleton on the on human inverse kinematics,” IEEE Transactions on Systems, mechanical loading of the low back in static holding tasks,” Man, and Cybernetics: Systems, vol. 49, no. 3, pp. 600–611, Journal of Biomechanics, vol. 83, pp. 97–103, 2019. [29] K. R. R. Holzbaur, W. M. Murray, G. E. Gold, and S. L. Delp, [15] L. H. Lu, Q. C. Wu, X. Chen, Z. Shao, B. Chen, and H. Wu, “Upper limb muscle volumes in adult subjects,” Journal of Bio- “Development of a sEMG-based torque estimation control mechanics, vol. 40, no. 4, pp. 742–749, 2007. strategy for a soft elbow exoskeleton,” Robotics and Autono- [30] M. Sakamaki, M. G. Bemben, and T. Abe, “Legs and trunk mous Systems, vol. 111, pp. 88–98, 2019. muscle hypertrophy following walk training with restricted [16] K. Y. Wu, Y. Y. Su, Y. L. Yu, C. H. Lin, and C. C. Lan, “A5- leg muscle blood flow,” Journal of Sports Science and Medicine, degrees-of-freedom lightweight elbow-wrist exoskeleton for vol. 10, no. 2, pp. 338–340, 2011. forearm fine-motion rehabilitation,” IEEE/ASME Transactions [31] X. Cui, W. H. Chen, X. Jin, and S. K. Agrawal, “Design of a 7- on Mechatronics, vol. 24, no. 6, pp. 2684–2695, 2019. DOF cable-driven arm exoskeleton (CAREX-7) and a control- [17] M. T. Picchiotti, E. B. Weston, G. G. Knapik, J. S. Dofour, and ler for dexterous motion training or assistance,” IEEE/ASME W. S. Marras, “Impact of two postural assist exoskeletons on Transactions on Mechatronics, vol. 22, no. 1, pp. 161–172, biomechanical loading of the lumbar spine,” Applied Ergo- nomics, vol. 75, pp. 1–7, 2019. [32] S. Głowiński, T. Krzyżyński, S. Pecolt, and I. Maciejewski, [18] K. Gui, U. X. Tan, H. Liu, and D. G. Zhang, “Electromyogra- “Design of motion trajectory of an arm exoskeleton,” Archive phy-driven progressive assist-as-needed control for lower limb of Applied Mechanics, vol. 85, no. 1, pp. 75–87, 2015. exoskeleton,” IEEE Transactions on Medical Robotics and Bionics, vol. 2, no. 1, pp. 50–58, 2020.

Journal

Applied Bionics and BiomechanicsHindawi Publishing Corporation

Published: May 25, 2021

References