Abstract
Prior research investigating human-exoskeleton interactions showed significant variabilities among users in their physical, physiological, and cognitive responses associated with exoskeleton use. Individual exoskeleton users’ varying anthropometric and demographic characteristics may lead to large inter-individual variability. This study examined the impacts of exoskeleton users’ body mass index (BMI), sex, and physical fitness on perceived exoskeleton fit, exoskeleton-related pain, and physical and cognitive task performances while wearing upper- and lower-body exoskeletons. The results indicated that participants with higher BMI were more likely to experience pain with an upper-body exoskeleton. Sex differences were also observed, with men experiencing more pain when using a lower-body exoskeleton. Frequent physical activity and cardio training were positively correlated with pain experienced with an upper-body exoskeleton. While the impacts were not universal across different exoskeleton types, the current findings offer insights into the potential causes of individual variability in human-exoskeleton interactions.
Introduction
Exoskeletons are mechanical structures made to support the strength and endurance of the user (Stirling et al., 2020). Exoskeletons are increasingly used in broad domains, including physical rehabilitation, industrial use, and warfighting, to assist users with various physically demanding tasks (Fox et al., 2019). This technology has the potential to enhance human capabilities; however, an improper fit may lead to users experiencing pain, obstructed movements, and increased rates of injury, resulting in ineffective use and adverse health consequences (Fox et al., 2019, Stirling et al., 2020). Users may also need to adopt different strategies to accommodate improper fitting (Stirling et al., 2020). Thus, the fit is likely to determine how effectively the users use an exoskeleton system to maximize its benefits.
Prior studies investigating human-exoskeleton interactions often showed substantial differences across individuals in the benefits associated with exoskeleton use. For example, exoskeleton users varied in muscle activation patterns and attentional and cognitive performance (Acosta-Sojo & Stirling, 2022; Afzal et al., 2018; Bequette et al., 2020; Leibman et al., 2022). Also, some users had greater impacts of exoskeleton use on their performance than others, and participants differed in having benefits or drawbacks from exoskeleton use (Afzal et al., 2018; Leibman et al., 2022).
Exoskeleton users may differ in their physical characteristics, such as the size of their limbs or bodies, which may determine their fit to exoskeletons. Exoskeletons are typically designed to be adjustable to fit varying body sizes (Stirling et al., 2020). However, exoskeleton designers may need to consider more comprehensive characteristics of a user’s body as exoskeletons could affect a wider bodily area as they add or redirect forces to even unsupported areas (Fox et al., 2019). The adjustability of an exoskeleton system may be difficult to meet the needs of all users, especially if users have extreme body sizes or unusual body shapes (Stirling et al., 2020). For example, some users with higher Body Mass Indexes (BMI), which indicates high body fatness, may have more issues fitting an exoskeleton properly due to difficulty finding the correct reference points. This may be particularly important in countries that have higher rates of population obesity, such as the United States (Flegal et al., 2012). Prior human-exoskeleton interaction studies have examined certain aspects of human anthropometry and exoskeleton design principles and identified key anthropometric parameters for exoskeleton design (Abhilash et al., 2021; Stirling et al., 2020). However, research investigating exoskeleton use by users with high BMI or unusual body sizes or shapes is limited (Abhilash et al., 2021; Gorgey, 2018).
Furthermore, there are differences in a variety of physical characteristics (e.g., bone structure, BMI, and body shape) between males and females, which may need to be considered when designing equipment (Fullenkamp et al., 2008; Wang et al., 2006). However, prior human-exoskeleton interaction studies have rarely examined sex differences in exoskeleton use performances, while others have relied on a pool of fit male subjects (e.g., Bequette et al., 2020). Previous studies suggest significant gender differences in exoskeleton use, with female and male users experiencing differential task demands, biomechanical forces, muscle loading, and perceived discomfort (Kim & Nussbaum, 2019; Park et al., 2022). For example, women showed increased peak hip flexion forces with exoskeleton use compared with non-exoskeleton use (Park et al., 2021).
Other factors may also affect human-exoskeleton interaction. For instance, the user’s physical fitness may impact their capabilities to control an exoskeleton system, as general physical fitness is positively associated with motor competence (Cattuzo et al., 2016). Users with regular physical training may be more adept at controlling an exoskeleton. However, there is a lack of research investigating how the user's general physical fitness level may impact the human-exoskeleton interaction.
The current study examined the impacts of body size, sex, and physical fitness on exoskeleton use, which have significant implications for understanding individual differences in exoskeleton use. Specifically, this study investigated five hypotheses: H1) users with higher BMI experience higher pain and poorer exoskeleton fit, H2) there are significant sex differences in exoskeleton fit and pain, H3) users with poorer physical fitness are more likely to experience pain and poorer fit, H4) the user’s overall exoskeleton fit is associated with pain experienced due to exoskeletons, and H5) poorer exoskeleton fit and pain resulted in greater changes in task performance with exoskeleton use.
Methods
Experimental Design
The current study was part of a larger study examining the effects of exoskeletons on physical and cognitive task performance. In this study, participants completed a dual task involving primary physical and secondary visual attention tasks. Participants were asked to move pegs through holes either around the shoulder or squatting height (overhead vs. squatting task conditions, respectively), each performed once with and once without the assistance of an exoskeleton (exo vs. non-exo conditions, respectively). Due to the significant disparity between the upper and lower body sample sizes, the data were separately analyzed.
Participants
A total of 41 students were recruited from an undergraduate student pool and compensated with class credit for participating in the study. All participants reported having normal or corrected to normal vision and no history of chronic pain in their shoulders, back, or legs. Only participants who fell within the recommended height range for the exoskeletons (165 – 190 cm) were recruited. Three participants were excluded from the study: two due to procedural errors and one due to data loss.
Among the final sample of 38, one participant was excluded from the data analysis for the overhead condition due to data loss with the display used for the visual attention task. Thus, the overhead task condition had a sample size of 37. For the squatting task, a substantial number of participants were excluded for several reasons. A total of 11 were excluded due to various exoskeleton-related errors that occurred during the preparation or task performance. An additional five were excluded because they could not fit properly into the lower body exoskeleton. Finally, one participant was excluded due to feeling pain during the squatting task. This resulted in the final sample size of 21 for the squatting condition. A summary of participant demographics for both the overhead and squatting conditions is presented in Table 1.
Participant demographics.
Materials and Stimuli
Exoskeletons
Two types of exoskeletons (Figure 1), a shoulder-based exoskeleton (Levitate Technologies Airframe) and a leg-based exoskeleton (SuitX LegX), were used for the overhead and squatting tasks, respectively. The Airframe exoskeleton provided mechanical assistance when participants held their arms above chest height. The LegX exoskeleton provided mechanical assistance when participants squatted.
Images of the lower (left) and upper (right) body exoskeletons used during the squatting and overhead tasks.
Physical Tasks
The two peg-in-hole tasks involved participants moving a peg into holes in a wooden plate affixed vertically against the wall. For the overhead peg-in-hole task, the plate was located 138cm above the ground at its center point, while the squatting peg-in-hole task was located 50cm above the ground.
Visual Detection Task
The visual detection task involved participants wearing an augmented reality (AR) head-mounted display that randomly presented targets across a visual field. Participants responded to a target’s appearance using a clicker which recorded the response and reaction time (RT).
Demographics Survey
The survey collected information on participants’ age, gender, height, and weight, as well as prior experience with wearable tools (e.g., a climbing harness and load belt) and VR/AR that may have impacted their performance in the experiment. The level of physical fitness was measured with the frequencies (i.e., the hours per week) of general activity, cardio training, and strength training. General activity was defined as non-dedicated training or activity, such as walking and physical labor at work. Cardio training was defined as explicit aerobic exercise such as running, cycling, swimming, etc., and strength training was defined as anaerobic training such as weightlifting.
Exoskeleton Pain and Fit Survey
The survey collected perceived fit and pain associated with exoskeleton use. The fit questions included four items asking about ease of fitting, comfort, anxiety, and obstructed movement on a five-point scale, with one being minimal negative experience (i.e., no anxiety, unobstructed movement, little donning effort) and five being greater negative impact (i.e., anxious, obstructed movement, extreme donning effort). The comfort question was an exception, with one being greater negative impact (i.e., uncomfortable) and five being minimal negative impact (i.e., comfortable). The comfort question was reverse-coded to match a higher fit score reflecting a worse experience with exoskeleton use. The pain questions asked about the levels of pain experienced during the exoskeleton use, on a 5-point scale, with a higher score being more intense pain. Participants reported the pain locations and rated the levels for each area.
Procedure
Participants were instructed to wear clothing suitable for fitting exoskeletons and physically demanding experiments. Once they arrived, participants gave informed consent to participate and then completed the demographics survey. Participants’ heights were measured using a stadiometer to confirm if they were within the acceptable height range for the exoskeletons. Participants were then briefed on the procedure for the following four dual-task blocks. Before beginning the experiment, the AR display was calibrated for each participant. They also went through a brief practice session to get used to the target display and the response clicker used in the visual attention task.
The two physical task conditions were done once with and without the use of an exoskeleton. The physical task involved participants completing a rotation, where they had to move a peg through eight holes in sequence. The experimenter measured the time to complete each sequence using a stopwatch. Participants were asked to complete as many rotations as possible in the 11-minute task period. While performing the peg-in-hole tasks, participants completed the secondary visual attention task concurrently. They were instructed to respond to the appearance of a visual stimulus on a handheld clicker as quickly as possible. The accuracy (i.e., detection rate) and RTs were recorded.
Prior to each experiment block utilizing an exoskeleton, participants were fitted to the exoskeleton with the aid of an experimenter. Participants had a practice session, during which they were asked to complete a series of motions to test the fit of the exoskeleton and to practice moving with the exoskeletons. The order of the four testing blocks (i.e., the overhead task with exo and non-exo and the squatting task with exo and non-exo) was partially counterbalanced. After each block using an exoskeleton, participants filled out the exoskeleton fit and comfort survey. Participants were instructed to leave the pain section blank if they did not feel any pain caused by the exoskeleton use. Participants also completed a workload questionnaire, which was not included in this paper as it is beyond the scope of the current study.
Results
Exoskeleton Fit and Pain
Self-reported exoskeleton fit and pain ratings are presented in Table 2. The participants generally experienced relatively positive fits and lower pain, and the fit and pain scores were higher (i.e., worse fit and pain) with the lower-body exoskeleton used during the squatting task than the upper-body exoskeleton used during the overhead task. Ten participants reported pain caused by the exoskeleton in at least one body part in both task conditions, and the mean pain level was calculated for each participant (i.e., the total scores of pains experienced by each participant divided by the number of pain areas). Most participants who reported pain caused by exoskeletons reported minimal pain, with scores ranging between 0 – 4 (Overhead M = 0.64, SD = 0.44; Squatting M = 0.78, SD = .48). Typically, pain was reported in the areas the exoskeleton and task involved: shoulders (n = 4) and arms (3) with the upper-body exoskeleton; and ankles (5) and knees (3) with the lower-body exoskeleton.
Self-reported exoskeleton fit and pain ratings during the overhead and squatting Tasks.
Effects of BMI on Exoskeleton Fit and Pain
BMI was calculated using the heights and weights and calculated using the following equation (Flegal et al., 2012):
The correlational analyses were conducted to examine the relationship between BMI and exoskeleton fit and pain ratings, separately for the overhead and squatting task conditions. A significant positive correlation was found between BMI and the pain ratings during the overhead task, r(37) = .387, p = .018, suggesting that participants with higher BMI experienced more severe pain when using the upper-body exoskeleton. Figure 2 presents the exoskeleton fit and pain ratings for the overhead condition across different BMI groups. No other associations were found to be significant (Tables 3 and 4).
Mean fit and pain ratings by BMI type during the overhead task. Note. Underweight n =1, Normal Weight n = 26, Overweight n = 6, Obese n = 3, Ex. Obese n = 2.
Correlations of BMI, fit and pain ratings, and task performance changes during the overhead task.
Significant at < .05 level.
Note. 1 physical task completion time; 2visual attention task accuracy and RT.
Correlations of BMI, fit and pain ratings, and task performance changes during the squatting task.
Significant at < .05 level.
Note. 1 physical task completion time; 2visual attention task accuracy and RT.
Sex Differences on Exoskeleton Fit and Pain
A pair of t-tests were conducted to compare the differences between males and females to determine if they differed in exoskeleton fit and pain. One participant was excluded because they did not identify their sex as male or female. For the overhead condition, there was no significant difference between males (fit M = 1.92, SD = 0.55; pain M = 0.28, SD = 0.43) and females (fit M = 1.67, SD = 0.60, pain M = 0.10, SD = 0.29) in the fit and pain ratings (fit: t(35) = 1.33, p = .193; pain: t(35) = 1.37, p = .137), indicating no sex-related differences on fit and pain when using the upper-body exoskeleton. For the squatting task, significant sex group differences were found in pain ratings, t(19) = 2.33, p = .031, d = .19. Male participants tended to experience higher pain (M = 0.62, SD = 0.64) than females (M = 0.14, SD = 0.22) with the lower-body exoskeleton. There was no significant difference between sexes (male M = 2.28, SD = 0.69; female M = 2.16, SD = 0.54) for the fit ratings, t(19) = 4.30, p = .672.
Effects of Physical Fitness on Exoskeleton Fit and Pain
Bivariate correlational analyses were conducted to examine the relationships between physical fitness and exoskeleton fit and pain ratings. Interestingly, a significant positive correlation was observed between pain with physical activity frequency, r(37) = .36, p = .030, and cardio training, r(37) = .34, p = .049, but not with strength training, r(37) = .29, p = .085, when reported for the overhead task. This suggests that participants who engaged in more physical activity and dedicated cardio training were more likely to experience higher levels of pain when using an upper-body exoskeleton. However, no similar associations were found for the squatting task: physical activity–pain: r(21) = -.25, p = .27; cardio–pain r(21) = .21, p = .355; strength–pain r(21) = -.10, p = .662, indicating that physical fitness was not associated with the lower-body exoskeleton pain. None of the physical fitness measures were associated with the upper- and lower-body exoskeletons, with r ranging between -.26 and .16.
Associations between Exoskeleton Fit and Pain
A significant correlation was found between exoskeleton fit and pain ratings during the squatting task, r(21) = .69, p = .003, suggesting that participants experiencing pain with the lower-body exoskeleton were more likely to report poorer fit. In particular, there was a significant relationship between the anxiety score and pain, r(37) = .69, p < .001, indicating that participants with more significant pain were more anxious while using the exoskeleton. However, participants’ self-reported fit was not significantly correlated with the pain ratings during the overhead task, r(37) = .30, p = .068. This may have been due to the small number of participants experiencing pain during the overhead task. However, among the four fit questions, the discomfort score was significantly correlated with pain, r(37) = .34, p = .042, which may suggest that participants who experienced more pain with upper-body exoskeleton use felt more discomfort.
Impacts of Fit and Pain on Task Performances
A correlational analysis was conducted to examine if exoskeleton fit and pain ratings were associated with greater changes in task performance with exoskeleton use. Three task performance measures were included: visual attention accuracy and RTs and physical task completion time. Changes in each task performance measure were calculated by subtracting the performance of the exo condition from the performance in the non-Exo condition. Positive scores indicate an increase in the measure from the non-exo to exo condition, and a negative score indicates a decrease. Thus, an increase in accuracy, a decrease in RTs, and a decrease in trial completion times indicate improvements in performance due to exoskeleton use. Conversely, decreased accuracy, increased RTs, and increased trial completion time would indicate degraded performance due to exoskeleton use. No significant correlation was found between fit and pain scores and changes in any of the task performance measures (Tables 3 and 4). This may suggest that poorer exoskeleton fit and pain do not necessarily impair task performance.
Discussion
This study examined if an exoskeleton user’s body size, sex, and physical fitness impact their interactions with an exoskeleton. The results suggest significant effects of the user’s body size measured using BMI on pain experienced while using an upper-body exoskeleton. Participants with higher BMI tended to experience greater pain with the upper-body exoskeleton during an overhead task. While no similar effect was found with the lower-body exoskeleton, it may suggest that exoskeletons could cause more severe pain among certain users with unusual body weights or shapes. Furthermore, the exoskeleton users’ pain and fit experiences differed between males and females, with males experiencing greater pain than females when using the lower-body exoskeleton. Our findings have implications for establishing evidence-based criteria or recommendations for exoskeleton fit to prevent user discomfort or pain.
This study also examined how regular physical fitness activities may impact human-exoskeleton interaction. Prior research highlighted that physical fitness could lead to improved task performance on motor skills due to higher motor competency (Cattuzo et al., 2016). However, our findings suggest that regular fitness activities could negatively impact exoskeleton use. Further investigation is warranted to identify the precise relations between physical fitness and human-exoskeleton interactions.
Prior research emphasizes that a good fit between the exoskeleton and the user is essential for effective exoskeleton use (Fox et al., 2019; Stirling et al., 2020). Our findings suggest that an exoskeleton user’s subjective fit experience may be critically impacted by pain associated with exoskeleton use. For instance, the results showed that users having pain with the upper-body exoskeleton were more likely to experience discomfort, suggesting a lack of pain is critical for positive human-exoskeleton interactions. However, the current study did not find that poorer fit and pain with exoskeleton use impacted the users’ physical and cognitive task performance when using exoskeletons. It is possible that the pain and improper fit experienced in the current experiment was relatively minimal; thus, it might not substantially impact the user’s exoskeleton use.
The current study had a few limitations. BMI scores are a simple bodily index calculated based on an individual’s height and weight and do not reflect more complex anthropometric characteristics. Specific measurements of segment size, muscle and fatty tissue mass, and other specific body measurements may need to be further investigated to establish the precise relations between body sizes and shapes and their impacts on human-exoskeleton interactions. Additionally, the current participants did not significantly vary in their BMI types, with most participants being in the normal weight group and only a few in other BMI groups (i.e., underweight, overweight, obese, extremely obese). Another limitation relates to the pain scores used in this study, which may not sensitively reflect participant experiences. It may be difficult for participants to precisely identify and verbalize the location, area affected, and intensity of pain on the self-reported pain survey. Due to the physical nature of the task, it is possible that participants may have misinterpreted pain caused by the exoskeleton and that of the task despite being instructed to report pain associated with exoskeleton use.
The findings suggest that exoskeleton designs should consider both anthropometric and demographic differences of potential users. Future studies should further examine potential factors that may contribute to human-exoskeleton interactions, including a broad range of anthropometric and demographic characteristics, such as varying body shapes, ages, and health and disability conditions.
