Analysis of Human–Exoskeleton System Interaction for Ergonomic Design

Abstract

Objective

Lower-limb exoskeleton systems are defined as gait training or walking-assisting devices for spinal cord injury or hemiplegic patients. Crutches, straps, and baffles are designed to protect subjects from falling. However, skin abrasions occur when the interaction forces are too large. In this study, the interaction forces between the human body and an exoskeleton system named the AIDER were measured to confirm whether the design was ergonomic.

Background

The AIDER system is a wearable lower-limb exoskeleton. It secures a subject by binding on the waist, thighs, shanks, and feet.

Method

Eight healthy subjects participated in the study. The interaction forces of the waist strap, thigh baffles, shank baffles, and crutch handles were measured by pressure sensors. Ten repetitions were completed in this study. After one repetition, custom comfort questionnaires were completed by the subjects.

Results

Although a few of the peak values of the maximum intensities of pressure between the hands and crutch handles reached the minimum value of the pain–pressure threshold (PPT), the average pressure intensities were much smaller than the PPT value.

Conclusions

The results indicated that the mechanical structure and control strategy of the AIDER must be improved to be more ergonomic in the future.

Keywords

exoskeleton AIDER system spinal cord injury interaction force questionnaires ergonomic

Introduction

According to the World Health Organization, as many as 500,000 patients suffer from spinal cord injuries (SCIs) every year worldwide (World Health Organization, 2013). People with SCIs are 2–5 times more likely to die prematurely, with worse survival rates in low- and middle-income countries (World Health Organization, 2013). Ordinary families cannot afford the cost of a SCI patient unless there is industrial injury compensation. Lifetime costs of a high tetraplegia patient can reach $74,509 per year (National Spinal Cord Injury Statistical Center [NSCISC], 2019). SCIs have a significant impact on quality of life (Behrman & Harkema, 2000) and health status. Dysfunction of the lower extremities is likely related to higher blood pressure, shorter life expectancy, social stigma, and increased rates of depression (Parent et al., 2011). Therefore, walking ability is usually cited as one of the primary goals of rehabilitation (Calhoun et al., 2013; Ditunno et al., 2008). An exoskeleton system is a typical system of human–machine interaction. It can assist paraplegia patients in walking around in their daily environments. Muscle strength and bone density seem to increase through gait training with use of exoskeleton systems (Karelis et al., 2017; Kolakowsky-Hayner et al., 2013; Raab et al., 2016). Therefore, exoskeleton systems are being widely used in rehabilitation (Suzuki et al., 2007).

Straps or baffles are used to connect subjects with exoskeleton systems. However, long-term excessive pressure can cause skin ulcers, skin tissue necrosis, and infection (Bass & Phillips, 2007; Karelis et al., 2017; Tamez-Duque et al., 2015). The static, dynamic, and shear forces produced by the interaction between the exoskeleton and human body should be within a certain safety threshold (Pons, 2008, 2010).

To ensure the security and comfort of an exoskeleton system, the interactive force between human and exoskeleton must be considered and analyzed (Kolakowsky-Hayner et al., 2013; Rathore et al., 2016). At present, there are three ways to measure the interaction force. (1) Pressure sensors can be used to measure the force directly (De Rossi et al., 2011; Lenzi et al., 2011; Tamez-Duque et al., 2015; Yandell & Zelik, 2016; Rathore et al., 2016). Researchers built a system that consisted of 20 pressure sensors and software, which could measure the distribution of the pressure in real time with fastening straps on the anterior thigh and anterior shank. This method could prevent the subjects from receiving skin injuries if excessive pressure was produced (Tamez-Duque et al., 2015). Other researchers constructed a real-time force measuring apparatus that consisted of 16 pressure sensors distributed on thigh and shank straps of an exoskeleton. The results indicated that the maximum pressure was on the anterior thigh straps and would cause skin injury if the exoskeleton were worn for a long time (Rathore et al., 2016). (2) The interactive force can be calculated using mathematical models. Cho et al. (2012) established a model of a human–machine system in the “AnyBody” software. The interactive forces on the subjects and exoskeleton could be calculated using the model (Cho et al., 2012). (3) Sensors can be used to measure the interaction forces and analyze biological signals (Hidler & Wall, 2005; Wilcox et al., 2016) using methods such as surface electromyography, electroencephalo-graph, and electro-oculogram.

However, most studies only focused on the interaction force between the shanks and shank baffles but ignored the interaction force between the hands and crutch handles. When walking with an exoskeleton with crutches (e.g., ReWalk and Ekso), subjects lean forward to maintain balance. Thus, the interaction force between the hands and crutch handles should not be ignored. In this paper, the subjects were required to complete various motions—sitting-to-standing, standing, walking, and standing-to-sitting—while wearing the AIDER (AssIstive Device for paRalyzed patients, University of Electronic Science and Technology, Chengdu, China) system. The interaction forces between their bodies and the straps/handles/baffles of the AIDER were measured. After the subjects completed the trial, the comfort evaluation questionnaire was required to be filled in. This study aimed to identify the safety and comfortability of the AIDER system and find problems with the design.

Method

Subjects

Eight healthy male subjects from the Center for Robotics of the University of Electronic Science and Technology of China participated in this study. They all signed the informed consent before the study. The subjects were 20–28 years old, their mean height was 174 ± 0.3 cm, and their mean weight was 63 ± 0.1 kg. There were not any female subjects in this study, because considerable muscle strength of the upper limbs was required when using AIDER. Generally, female muscle strength is lower than male muscle strength.

All of the subjects had no physical disabilities, history of encephalopathy, cognitive dysfunction, or hearing disorders. This study was approved by the Ethics Committee of the Sichuan Provincial Rehabilitation Hospital. All of the subjects signed informed consent forms before the tests.

Devices

The AIDER system was used for this study, which is a lower-limb wearable exoskeleton system designed to help SCI patients with walking whose injury planes are between T5 and L1. The total weight was 23 kg. The computer was in a box on the subject’s lower back, and its weight was approximately 4 kg. The AIDER system has 8 degrees of freedom (DOFs), 2 in the hip joint, 1 in the knee joint, and 1 in the ankle joint. The hip joint and knee joint had active drives, and four motors fit on each hip joint and each knee joint (Yue et al., 2017). The ankle joint had one passive drive, with a spring fit on each ankle joint (Yue et al., 2017). The AIDER system was equipped with a pair of crutches to control the motion and maintain balance (Hassan et al., 2014). As shown in Figure 1, the subject's waist, hands, thighs, shanks, and feet contacted the AIDER via handles, straps, or baffles.

Figure 1

AIDER system.

The pressure sensors (FlexiForce A201, USA) were used to measure the interaction forces. The diameter of the surface of one sensor was 9.53 mm. There were three sensors on each crutch handle and nine on each shank baffle. The sampling frequency of each sensor was 20 Hz, and the sensor could measure forces in the 0–110 N range.

Experimental Procedure

During a gait period, the standing and swing phases were distinguished from each other based on the marker trajectories on the feet. After a training session, comfort questionnaires were completed by the subjects. Before the study, 1 day was used to train the subjects how to use the AIDER system. The session included training on sitting-to-standing, standing, maintaining balance, walking for 10 m, and standing-to-sitting. Figure 2 shows a complete gait period using the AIDER for walking. The gray lines represent the crutches of the AIDER, and the red and blue lines represent the right and left side of the subject’s body, respectively. The subjects learn how to use the AIDER to walk following Steps 1–8.

Figure 2

Detailed training process.

Data Collection and Analysis

The interaction forces of the waist strap, thigh baffles, shank baffles, and crutch handles were collected, which directly contacted the subjects’ bodies, as shown in Figure 3. The interaction forces between the body and waist baffles, crutch handles, shank baffles, and thigh baffles were measured and compared with the pain–pressure threshold (PPT) value (Fransson-Hall & Kilbom, 1993; Lee et al., 2005; Rocon et al., 2008; Rolke et al., 2005). If the interaction force exceeded the PPT value, the design of the AIDER was considered to be unreasonable and unsafe. In addition to measuring the interaction force, comfort questionnaires were also used to assess the comfort of the AIDER.

Figure 3

Sensor layout.

The positions of all the sensors used in this study are shown in Figure 3. Figure 4 shows the detailed sensor layout of the crutch handles and shank baffles. The positions of the sensors were fixed in the same location for each day of data collection. The subjects were required to wear the AIDER and complete 10 repetitions of sitting-to-standing, standing for 10 min, walking for 10 m, and standing-to-sitting. After one repetition was completed, the subjects were required to finish the questionnaire. These 10 repetitions did not need to be completed in 1 day. However, each subject was required to finish the trial within 3 days. The data collection was completed in 1 month. The comfort evaluation questionnaire was self-made to evaluate the comfortability of AIDER.

Figure 4

Detailed sensor layout of crutch handles and shank baffles.

IBM SPSS Statistics 22 was used to conduct statistical analysis. We used the independent t test to analyze the statistical differences between the pressure of the left and right shank baffles/crutch handles. The Kruskal–Wallis test is a nonparametric statistical test that is used to assess the differences between three or more independently sampled groups of a single, nonnormally distributed continuous variable (McKight & Najab, 2010). Therefore, the Kruskal–Wallis test was used to determine whether there was a statistical difference between the interaction forces of different subjects and the 10 repetitions of each subject. The significance level was set to 0.05.

Results

The pain threshold is the body’s maximum tolerance to external stimuli. Every part of the human body can tolerate different levels of pain. The maximum intensities of pressure on the crutch handles and shank baffles of all the repetitions were compared with the corresponding PPT value (Figure 5). The differences in the intensities of the pressure between the left and right baffles (or straps, handles) were tested by the independent t test with a significance level of 0.05 for comparison. The result was 0.000, which indicated that the interaction forces of left and right sides were significantly statistically different. The Kruskal–Wallis test was used to assess the differences in the interaction pressures between the 8 subjects and the 10 repetitions of 1 subject with a level of significance as 0.05. The test result of the interaction forces of different subjects was 0.000, which indicated that the forces were significantly different for the eight subjects. This indicated that the interaction forces of the different subjects varied. Furthermore, the Kruskal–Wallis test result for the 10 repetitions of Subject 01 was 0.086, which indicated that there was no significant difference between the 10 repetitions. The Kruskal–Wallis test results for Subjects 02–08 were 0.075, 0.211, 0.130, 0.146, 0.870, 0.749, and 0.662, respectively. Therefore, we chose one repetition for each subject for the analyses presented in Sections “Sitting-to-Standing,” “Standing,” “Walking,” and “Standing-to-Sitting.” Figures 6 -undefined show the interaction force between the subjects’ hands and crutch handles, shanks, and shank baffles when sitting-to-standing, standing, walking, and standing-to-sitting. To analyze the interaction force, the time was normalized in each phase.

Figure 5

Maximum intensity of pressure.

Figure 6

Interaction forces of the crutch handles and shank baffles when sitting-to-standing.

Figure 7

Interaction force of crutch handles and shank baffles when standing.

Figure 8

Interaction force of crutch handles and shank baffles when walking.

Figure 9

Interaction forces of crutch handles and shank baffles when standing-to-sitting.

Comfort Questionnaires

In this trial, the self-made questionnaire for comfort evaluation of the AIDER was used. When the subjects completed sitting-to-standing, standing, walking, and standing-to-sitting, they were required to complete the questionnaire immediately. Table 1 shows the result of this questionnaire. The number in the table represents the number of subjects that selected this option. The results indicated that all of the subjects agreed that using, wearing, and adjusting the size of AIDER was easy. Furthermore, they all understood how to operate AIDER with one demonstration. Three subjects felt uncomfortable when wearing the AIDER, and the other three felt neutral. However, when using the AIDER, four subjects felt slightly uncomfortable and one felt much more uncomfortable. Most of the subjects did not feel tired and trusted the AIDER when using it.

Table 1

Comfort Evaluation Questionnaire for AIDER

Illustration	Evaluation
Illustration	Strongly Agree	Agree	Neutrality	Disagree	Strongly Disagree
a) It was easy to use the exoskeleton	100%
b) It was easy to wear the exoskeleton and adjust its size	75%	25%
c) It was comfortable to wear the exoskeleton		25%	37.5%	37.5%
d) Using the exoskeleton was not too painful		50%	25%	25%
e) I did not feel tired when walking with the exoskeleton	25%	50%	12.5%	12.5%
f) I felt comfortable when using the exoskeleton			37.5%	50%	12.5%
g) I trusted the exoskeleton when using it and believed no accident would occur	62.5%	25%	12.5%

Maximum Intensity of Pressure

The maximum intensity of pressure of the 10 repetitions is shown in Figure 5. To compare with the PPT value of the corresponding part, the left three boxes in each figure represent the intensity of the pressure of the left hand, right hand, and the PPT value of hand, and the right three boxes in each figure represent the intensity of the pressure of the left shank, right shank, and the PPT value of shank. Since the PPT value is affected by age (Lee et al., 2005), we chose the value for males between 20 and 28 years. Figure 5a shows the maximum intensity of pressure when sitting-to-standing. The peak value of the left hand was 474 kPa, which nearly reached the threshold of the PPT of the hand (482 kPa). The intensity of the pressure of the shank was far from the PPT value of the shank. Although most of the data did not reach the threshold of the PPT value, the intensity of the pressure of the hand was still too large to tolerate. It may not have caused skin abrasions, but the subjects felt uncomfortable. Figure 5b shows the maximum intensity when standing. It did not reach the threshold of the PPT value for the hands and shanks. The subjects stood with the AIDER easily and without discomfort. Standing was a static phase. The subjects leaned forward slightly to maintain balance, and the pressure on the crutch handles and shank baffles did not exceed the threshold. When the subjects were walking, the intensity of the pressure increased (Figure 5c). The peak value of the intensity of pressure of the left hand (504 kPa) and right hand (505 kPa) exceeded the minimum PPT value of the hand (482 kPa). However, most of the data were below the threshold of the PPT value. When the subjects were standing-to-sitting, more force was needed to support their bodies using the crutches. The maximum intensity of pressure is shown in Figure 5d. The peak value of the left hand reached 514 kPa, which exceeded the threshold of the PPT value by the largest amount of all of the phases. Although most of the data focused below the PPT value, the intensity of the pressure was still too large to tolerate. This could cause the subjects discomfort if this state were maintained for too long. The intensities of the pressure of the shanks were much smaller than the PPT value of the shanks for all of the motions.

Sitting-to-Standing

As shown in Figure 6, during the sitting-to-standing phase, the subjects first put the crutches behind the chair to support their bodies. They subsequently gradually moved the crutches forward to stand up, and their knees transitioned from a 90° (sitting) angle to 180° (standing). Because the time of sitting-to-standing was different for each subject, it was normalized in Figure 6. During each time block, the range of the average interaction force for all of the subjects is indicated by boxes.

Figure 6a and b shows the average interaction forces between the hands and the crutch handles. The average maximum interaction force appeared in the beginning. While the subjects were standing up, the force between the hands and crutch handles decreased. When the subjects began to stand up, they shifted almost all their weight to the crutch handles to support their bodies. Therefore, the interaction force between the hands and crutch handles increased during standing.

Figure 6c and d shows the average interaction forces between the shanks and shank baffles. The average interaction force decreased with time. The reason was that when the subjects sat on the chair, their shanks and the shank baffles did not interact too much. When the subjects stood up, their shanks and shank baffles were moved closer together and more interaction forces were produced. When they completed the motion, the interaction force became stable.

Standing

The standing phase began when the subjects stood upright with the AIDER. In this phase, the subjects were required to stand and maintain balance for 10 min with their knees kept at 180°, as shown in Figure 7. We believe that 10 min of standing was sufficient to obtain stable data to obtain the maximum force. In addition, we observed whether skin abrasions occurred. If the time were too short, convincing conclusions could not be obtained, and if the time were too long, the subjects would become impatient.

As shown in Figure 7a and b, the average interaction force between the subjects’ hands and the crutch handles remained stable during standing. When the subjects stood upright with the AIDER, they did not need the crutches to support their bodies. Therefore, the average interaction force remained stable.

Figure 7c and d shows the tendency of the average interaction force between the shanks and shank baffles. There was no increasing or decreasing tendency during standing. While the subjects remained standing, they did not rely on the shank baffles for support.

Walking

The walking phase required the subjects to walk for 10 m. The first step always began with the right leg which was set by the AIDER system. When the right leg took a step forward and landed, the subject moved the right crutch forward and prepared for the second step with the left leg. This process was repeated until the subject walked 10 m.

Figure 8a and b shows the average interaction forces between the hands and crutch handles. In the walking phase, the subjects needed to shift their center of gravity to the left as the right leg prepared to take a step, and vice versa. Therefore, the average force moved up and down from the beginning to the end. Furthermore, the left and right sides had opposite tendencies. When the left side reached its maximum value, the right side reached a minimum.

The average interaction force between the shanks and shank baffles (Figure 8c and d) did not show evident tendencies. While the subjects were walking, the average force floated between 0 and 60 N.

Standing-to-Sitting

The standing-to-sitting phase was the opposite motion to the sitting-to-standing phase. The subject moved the crutches behind gradually and sat down slowly, with their knees bending from 180° to 90°.

Figure 9a and b shows the interaction forces between the hands and the crutch handles. The forces were greater in the middle of the motion than at the beginning and end. This indicated that the crutches provided considerable bracing forces once the crutches began moving.

As shown in Figure 9c and d, the average interaction force between the shanks and shank baffles did not vary significantly because the subjects did not rely on the shank baffles during the standing-to-sitting motion.

Discussions

In addition to the crutch handles and shank baffles, the interaction forces of the waist straps and thigh baffles were also measured, but the force values were nearly 0 N. One possible reason was that the healthy subjects could stand up or walk by themselves and did not touch the baffles. Although they tried their best not to use their own lower extremities to move, they were never able to act completely as paraplegic patients. The second reason might be that the designs of the waist straps and thigh baffles were inadequate. When the subjects were standing, walking, or sitting, the waist straps and thigh baffles did not provide bracing forces for their bodies. The healthy subjects could stand, walk, or sit by themselves, but SCI subjects cannot. The most likely reason was that the interaction force of the waist straps and thigh baffles could not be measured precisely by the pressure sensors. They could measure the vertical force exactly, but the accuracy of the measured tangential force was not guaranteed. The design of the waist straps and thigh baffles should be studied further in the future.

According to Figure 5 and the analysis in Section “Maximum Intensity of Pressure”, statistical differences of the maximum force on the left and right crutch handles or shank baffles were not found. This indicated that the subjects were not leaning left or right. Furthermore, the maximum intensities of pressure on the crutch handles were greater than those of the shank baffles. A few of the peak values of the maximum intensity of pressure of the left hand exceeded the PPT value (Lee et al., 2005; Rocon et al., 2008; Rolke et al., 2005) when sitting-to-standing and standing-to-sitting. When the subjects were walking, a few of the maximum intensities of the pressure on the two hands exceeded the PPT value. This may have been the reason the subjects felt uncomfortable when using the AIDER, which is shown in the comfort questionnaire in Table 1. Furthermore, the maximum intensities of pressure on the crutch handles when sitting-to-standing, walking, and standing-to-sitting were greater than that while standing. The reason was that the subjects put almost all of their weight on the crutch handles when standing up, walking, and sitting down, whereas they could maintain their balance when standing without relying on the AIDER. Furthermore, if the subjects felt numb or uncomfortable in their hands, this was possibly because the contact area between the hands and crutch handles was not large enough, and the design was not ergonomic. Therefore, the design of the crutch handle and shank baffles must be modified to be more ergonomic.

The average interaction force of each subject was analyzed based on Figures 6 -undefined. Ten repetitions were conducted by each subject in this study, and the data did not show significant differences. Therefore, one repetition of each subject was selected for analysis. When the subjects were standing up, the average interaction force between the hands and the crutch handles tended to decrease and fluctuated throughout the phase, because the subjects needed the crutches to support their bodies in the beginning and did not need it as much when they were standing almost upright. Furthermore, the average interaction forces between the shanks and shank baffles increased from the initial forces. The increasing tendency indicated that when the subjects were standing up, their shanks had closer contact with the shank baffles than when sitting, and consequently, the interaction force increased. When the subjects were standing and maintain their balance, the average interaction force did not oscillate much at the beginning and in the middle of the phase. During the walking phase, the interaction force rose and fell from the beginning to the end. Furthermore, the average interaction force between the left and right sides has opposite tendencies. The reason was that the subjects needed to shift their centers of gravity when walking. When the subjects were sitting, the curve of the interaction force with the hands began to oscillate in the middle of the motion. The reason was that, when the subjects began to sit down, they moved the crutches behind and put most of their body weight on the crutches until they sat down.

For patients with the SCI plane under T1 (Kirshblum et al., 2011), their upper-limb pain tolerances were almost the same as those of healthy people because their upper limbs were not injured. If patients suffered from complete SCI, their lower limbs did not have proprioception, which included pain perception. Patients with incomplete SCI also experienced a decrease in the perception of suprathreshold harmful stimuli.

In this version of the AIDER system, the control button was on the crutch handles. The subjects should use the crutches to support their body weights while pressing the control button, which may cause the interaction force to exceed the PPT value. This is one of the reasons that some subjects reported that their hands were uncomfortable. If the control button were replaced by posture recognition or another control method, the subjects might feel less discomfort when using the AIDER system. Furthermore, the mechanical structure of the whole system should be modified to be more ergonomic. The subjects in this study all leaned forward when standing and walking, which is not a normal posture. This may have created more interaction forces on their hands and shanks and small interaction forces between their thighs and the thigh baffles.

Limitations

The subjects of this study were not paraplegic patients, because the safety of the AIDER had not been verified before this study. Skin abrasions may occur when the interaction force exceeds the PPT value (Fransson-Hall & Kilbom, 1993; Lee et al., 2005; Rocon et al., 2008; Rolke et al., 2005), which means excessive pressure may cause skin sores, festering, or more severe injuries. If the paraplegic patients were injured again, their health conditions will be more severe. Secondary injury may cause permanent damage of their body functions. Because their blood circulation is slower than noninjured people, it is difficult for them to recover from even minor bruises. Therefore, only healthy subjects participated in this study. Although the data of the healthy subjects are likely different from that of actual patients, it can indicate problems with the design of the AIDER. After improving the design of the AIDER, paraplegic patients can participate in the subsequent studies of the clinical trials.

Female subjects did not participate in this study because considerable upper-limb strength may be needed when using the AIDER system. Typically, males have more strength than females. Furthermore, skin tolerance is different between females and males. Therefore, male subjects were chosen. In future studies, the differences of the interaction forces for females and males when using the AIDER system will be analyzed.

Conclusions

According to the analysis, the design of the waist straps and thigh baffles were not ergonomic, because they played no role in supporting the body. Although a few of the peak values of maximum pressure intensity between the hands and crutch handles reached the minimum value of the PPT (Lee et al., 2005; Rocon et al., 2008; Rolke et al., 2005), the average intensities of the pressure on the hands and shanks were much smaller than the PPT value. However, the interaction forces between the hands and crutch handles caused most of the subjects to feel discomfort in their hands during the motion, according to their questionnaires or oral reports. This indicated that the mechanical structure and control strategy of the AIDER must be made more ergonomic in future designs.

Key Points

Crutches, straps, and baffles were designed to protect the subjects from falling. However, skin abrasions can occur if the interaction force is too large.

The interaction forces between human body and the AIDER system were measured to confirm whether the design was ergonomic.

The results showed that the maximum intensity of pressure approached the PPT value. Although the peak value did not last long, it indicated that the risks of skin abrasions were present. In future designs, more attention will be paid to ergonomics.

Footnotes

Acknowledgments

This research project was supported by the National Key Research and Development Plan (2017YFB1302300) and the Fundamental Research Funds for the Central Universities (A03013023801147). We thank Accdon (www.accdon.com) for its linguistic assistance during the preparation of this manuscript. All authors are employed by University of Electronic Science and Technology of China, and their research and authorship of this article was completed within the scope of their employment with University of Electronic Science and Technology of China.

ORCID iD

Jing Qiu

Author Biographies

Yilin Wang is a doctoral candidate in the School of Automation Engineering at the University of Electronic Science and Technology of China, Chengdu, Sichuan Province, China. She is interested in exoskeleton systems.

Jing Qiu is an associate professor in the School of Mechanical and Electrical Engineering, University of Electronic Science and Technology of China, Chengdu, Sichuan Province, China. She received her PhD in ergonomics from the Technical University of Darmstadt in 2010.

Hong Cheng is a professor in the School of Automation Engineering at the University of Electronic Science and Technology of China, Chengdu, Sichuan Province, China. He received his PhD in control science and engineering from Xi’an Jiaotong University in 2003.

Xiaojuan Zheng received her master degree in mechanical design and automation from the University of Electronic Science and Technology of China in 2018. She is now working on ergonomics.

References

Bass

M. J.

Phillips

L. G

. (2007). Pressure sores. Current Problems in Surgery, 44, 101–143. https://doi.org/10.1067/j.cpsurg.2006.12.007

Behrman

A. L.

Harkema

S. J

. (2000). Locomotor training after human spinal cord injury: A series of case studies. Physical Therapy, 80, 688–700. https://doi.org/10.1093/ptj/80.7.688

Calhoun

C. L.

Schottler

Vogel

L. C

. (2013). Recommendations for mobility in children with spinal cord injury. Topics in Spinal Cord Injury Rehabilitation, 19, 142–151. https://doi.org/10.1310/sci1902-142

Cho

Kim

Jung

Lee

. (2012). Analysis and evaluation of a combined human-exoskeleton model under two different constraints condition. Proceedings of the International Summit on Human Simulation, St. Pete Beach, FL, USA. 23–25.

De Rossi

S. M. M.

Vitiello

Lenzi

Ronsse

Koopman

Persichetti

Vecchi

Ijspeert

A. J.

van der Kooij

Carrozza

M. C

. (2011). Sensing pressure distribution on a lower-limb exoskeleton physical human-machine interface. Sensors, 11, 207–227. https://doi.org/10.3390/s110100207

Ditunno

P. L.

Patrick

Stineman

Ditunno

J. F

. (2008). Who wants to walk? Preferences for recovery after SCI: A longitudinal and cross-sectional study. Spinal Cord, 46, 500–506. https://doi.org/10.1038/sj.sc.3102172

Fransson-Hall

Kilbom

. (1993). Sensitivity of the hand to surface pressure. Applied Ergonomics, 24, 181–189. https://doi.org/10.1016/0003-6870(93)90006-U

Hassan

Kadone

Suzuki

Sankai

. (2014). Wearable gait measurement system with an instrumented cane for exoskeleton control. Sensors, 14, 1705–1722. https://doi.org/10.3390/s140101705

Hidler

J. M.

Wall

A. E

. (2005). Alterations in muscle activation patterns during robotic-assisted walking. Clinical Biomechanics, 20, 184–193. https://doi.org/10.1016/j.clinbiomech.2004.09.016

10.

Karelis

A. D.

Carvalho

L. P.

Castillo

M. J.

Gagnon

D. H.

Aubertin-Leheudre

. (2017). Effect on body composition and bone mineral density of walking with a robotic exoskeleton in adults with chronic spinal cord injury. Journal of Rehabilitation Medicine, 49, 84–87. https://doi.org/10.2340/16501977-2173

11.

Kirshblum

S. C.

Burns

S. P.

Biering-Sorensen

Donovan

Graves

D. E.

Jha

Johansen

Jones

Krassioukov

Mulcahey

M. J.

Schmidt-Read

Waring

. (2011). International standards for neurological classification of spinal cord injury (revised 2011). The Journal of Spinal Cord Medicine, 34, 535–546. https://doi.org/10.1179/204577211X13207446293695

12.

Kolakowsky-Hayner

S. A.

Crew

Moran

Shah

. (2013). Safety and feasibility of using the Ekso™ bionic exoskeleton to aid ambulation after spinal cord injury. Journal of Spine, 4.

13.

Lee

W. C.

Zhang

Mak

A. F

. (2005). Regional differences in pain threshold and tolerance of the transtibial residual limb: Including the effects of age and interface material. Archives of Physical Medicine and Rehabilitation, 86, 641–649. https://doi.org/10.1016/j.apmr.2004.08.005

14.

Lenzi

Vitiello

Rossi

S. M. M. D.

Persichetti

Giovacchini

Roccella

Vecchi

Carrozza

M. C

. (2011). Measuring human–robot interaction on wearable robots: A distributed approach. Mechatronics, 21, 1123–1131. https://doi.org/10.1016/j.mechatronics.2011.04.003

15.

McKight

P. E.

Najab

. (2010). Kruskal‐Wallis test. The Corsini Encyclopedia of Psychology, 1.

16.

National Spinal Cord Injury Statistical Center (NSCISC) . (2019). Facts and figures at a glance. University of Alabama at Birmingham. Retrieved April 26, 2019, from https://www.nscisc.uab.edu/

17.

Parent

Mac-Thiong

J.-M.

Roy-Beaudry

Sosa

J. F.

Labelle

. (2011). Spinal cord injury in the pediatric population: A systematic review of the literature. Journal of Neurotrauma, 28, 1515–1524. https://doi.org/10.1089/neu.2009.1153

18.

Pons

J. L

. (2008). Wearable robots: Biomechatronic exoskeletons. John Wiley & Sons.

19.

Pons

J. L

. (2010). Rehabilitation exoskeletal robotics. IEEE Engineering in Medicine and Biology Magazine, 29, 57–63. https://doi.org/10.1109/MEMB.2010.936548

20.

Raab

Krakow

Tripp

Jung

. (2016). Effects of training with the ReWalk exoskeleton on quality of life in incomplete spinal cord injury: A single case study. Spinal Cord Series and Cases, 2, 15025. https://doi.org/10.1038/scsandc.2015.25

21.

Rathore

Wilcox

Ramirez

D. Z. M.

Loureiro

Carlson

. (2016, August). Quantifying the human-robot interaction forces between a lower limb exoskeleton and healthy users [Conference session]. In 2016 38th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), IEEE. 586–589.

22.

Rocon

Ruiz

A. F.

Raya

Schiele

Pons

J. L.

Belda‐Lois

J. M.

Poveda

Vivas

M. J.

Moreno

J. C

. (2008). Human–robot physical interaction. In Wearable robots: Biomechatronic exoskeletons (pp. 127–163). John Wiley & Sons, Ltd.

23.

Rolke

Andrews Campbell

Magerl

Treede

R.-D

. (2005). Deep pain thresholds in the distal limbs of healthy human subjects. European Journal of Pain, 9, 39–48. https://doi.org/10.1016/j.ejpain.2004.04.001

24.

Suzuki

Mito

Kawamoto

Hasegawa

Sankai

. (2007). Intention-based walking support for paraplegia patients with robot suit HAL. Advanced Robotics, 21, 1441–1469.

25.

Tamez-Duque

Cobian-Ugalde

Kilicarslan

Venkatakrishnan

Soto

Contreras-Vidal

J. L

. (2015). Real-time strap pressure sensor system for powered exoskeletons. Sensors, 15, 4550–4563. https://doi.org/10.3390/s150204550

26.

Wilcox

Rathore

Morgado Ramirez

D. Z.

Loureiro

R. C. V.

Carlson

. (2016). Muscular activity and physical interaction forces during lower limb exoskeleton use. Healthcare Technology Letters, 3, 273–279. https://doi.org/10.1049/htl.2016.0063

27.

World Health Organization . (2013). Spinal cord injury: As many as 500 000 people suffer each year. World Health Organization. Retrieved March 20, 2019, from https://www.who.int/mediacentre/news/releases/2013/spinal-cord-injury-20131202/en/

28.

Yandell

M. B.

Zelik

K. E

. (2016). Transforming how we physically integrate exoskeletons [Paper presentation]. 40th Annual Meeting of the American Society of Biomechanics, USA, Raleigh.

29.

Yue

Cheng

Chen

Deng

Lin

. (2017, August). Design of a wearable sensing system for a lower extremity exoskeleton [Conference session]. In 2017 IEEE International Conference on Mechatronics and Automation (ICMA). China, Hong Kong.464–468.