Effects of exoskeleton use on movement kinematics during performance of common work tasks: A case study

Abstract

BACKGROUND:

An exoskeleton may assist performance of basic work-related tasks. Its application should not alter user kinematics, which compromise user safety.

OBJECTIVE:

This case study was used to assess whether people wearing a lower-body K-SRD^TM exoskeleton could complete common work tasks without altering kinematics that may increase injury risk.

METHODS:

Three males performed three tasks: kneeling and standing (kneel), lifting and lowering a weighted box floor-to-waist (lift), and stair-climbing with a weighted box (climb), all repeated with and without exoskeleton use (EXO, NONE).

RESULTS:

Kinematics with EXO often mimicked NONE. Hip and knee flexion with EXO often exceeded NONE without increasing heart rate for kneel. During lift with EXO, participants avoided greater lateral trunk flexion associated with injuries and used the preferred semi-squat technique. Participants produced more foot clearance with EXO than NONE during climb. Other outcomes of heart rate, perceived exertion, fatigue, and usability were mixed.

CONCLUSIONS:

EXO augmentation does not need to alter movement kinematics during performances of kneel, lift, and climb tasks. EXO kinematic alterations did not appear to compromise user safety in terms of lateral trunk bending. It may encourage good technique, such as greater foot clearance to avoid tripping, for some tasks, and changes in lifting strategies to avoid extreme flexion and protect passive tissues.

Keywords

Ergonomics human performance augmentation injury prevention musculoskeletal disorders

1. Introduction

Work-related musculoskeletal injuries are costly in terms of personal and family suffering and in terms of economic productivity to the employer and the nation’s economy. Unfortunately, workers from many industrial trades are afflicted with these injuries. Using construction as an example, injuries to the back and lower extremities account for approximately 55% of work-related musculoskeletal disorders (WMSDs) [1]. Self-reported prevalence of back pain ranges from 26–60% and prevalence of lower limb pain ranges from 31–44% [1 –3]. Over 70% of WMSDs are caused by overexertion from lifting or other activities [1]. Injuries to the back and lower limb often result from similar activities, which involve stooping or twisted postures and heavy lifting [4]. This may suggest that workers in several industries performing similar tasks are at risk.

Although WMSD risk for the back and lower extremities is high, effective interventions are not always well-integrated into work practices [2, 5]. Misuse of “proper” lifting techniques can and often do occur, not only because of the frequently changing work settings [6], but also because of fatigue [7]. Highly variable environments, like those in industrial settings, prevent eliminating or minimizing safety hazards through engineering controls, such as mechanical lifting devices for manual materials handling (MMH) tasks, which could reduce physical work demands and lower risk of low back disorders [8, 9]. Moreover, financial concerns, including initial costs and productivity changes, drive industry adoption of any needed or required controls and may provide a bigger incentive for implementation than reduced health risks [5, 10]. Therefore, successful interventions must meet cost and time requirements in addition to reducing WMSD risk.

Given the need for workers to observe and adapt to their inconsistent surroundings, exploitation of human-robot collaboration (HRC) provides an interesting solution. One HRC paradigm involves the use of exoskeletons, which match the mobility of the user and reduce WMSD risk, simultaneously. Exoskeletons are robotic devices that attach to the human body externally for the purpose of delivering mechanical power to augment or replace user strength, endurance, and/or mobility. One category of exoskeletons assists with lower limb functions such as sitting down, standing up, walking, balancing, kneeling, and squatting. This type of human performance augmentation (HPA) exoskeleton can increase the physical capabilities of able-bodied individuals [11]. Assessing state-of-the-art of exoskeletons is inherently complex due to the rapidly evolving nature of the field [11], where announcements of new devices occur quite frequently. Also, given the developmental stage of the field, competition among organizations, and the high financial stakes, developers who protect their intellectual property offer few peer-reviewed publications [11].

While current research and development in the field of HPA exoskeletons mainly focuses on military and medical applications, exoskeletons have the potential for use as an effective intervention for MMH and other physically demanding industrial tasks. In fact, the fastest growing field of exoskeleton research involves wearable robotics designed for industrial use. Potential benefits of this technology are: (1) reduction of work-related injuries, cutting costs associated with medical fees, sick leave, and lawsuits; (2) reduction of worker fatigue and increase of worker alertness, productivity, and work quality; and (3) keeping skilled and experienced personnel in the workforce longer [12]. Potential pitfalls of the technology include: (1) localized discomfort, increased compensatory muscle activity [13], potential muscle atrophy, and unsafe changes in movement kinematics for the wearer; (2) affordability, battery life, and weight of the device; and (3) extended time to become proficient with the device [14]. Newer, more evolved exoskeletal technology seems to accompany greater potential benefits and fewer potential drawbacks. Researchers need to determine which benefits and pitfalls exist for a given device and whether the potential benefits outweigh the potential pitfalls of each exoskeleton.

As of today, no established standards for evaluating and quantifying the performance of HPA exoskeletons exist. Researchers commonly determine the metabolic cost of performing a task with and without exoskeleton augmentation to assess if assistance provided by the device can compensate for the added weight to the lower body and the alterations in efficient gait dynamics of users. Few studies show a meaningful change in metabolic cost for HPA exoskeletons used for locomotion [11]. Tasks such as lifting objects, squatting, kneeling, rising, and/or climbing make it more difficult to assess the quality of movement required for safety and productivity. To address this, the National Institute of Standards and Technology (NIST) recently hosted a meeting of government agencies “to discuss the current state of standards development within the wearable robotics space and to identify gaps in standards for safety, performance, interoperability, ergonomics, and cybersecurity” [15]. With a long range goal to help in this endeavor by seeking to introduce standards for exoskeleton use during task performances used in industry, the purpose of this case study was to present preliminary findings on whether a lower body HPA exoskeleton will modify user movements during performance of ordinary tasks. Specifically, it was to determine if young healthy people wearing a lower-body K-SRD^TM exoskeleton (B-TEMIA Inc, Fig. 1), originally designed for military applications, could perform basic work-related tasks without altering kinematics that may increase risk for WMSDs.

Fig. 1.

K-SRD^TM exoskeleton from lateral (left) and posterior (right) views. The backpack observed in the posterior view contains the system’s battery.

2. Materials and methods

2.1. Participants

Three male participants (age: 19–22 years; height: 165–188 cm; mass: 61–106 kg) read the procedures and signed the informed consent approved by the Louisiana State University Institutional Review Board before participating in lab-based tasks mimicking typical manual labor work activities with and without exoskeleton augmentation. Participants either performed manual labor tasks at their job (P1) or exercised regularly (P2, P3) with no history of current musculoskeletal or other health disorders that may have impacted their performance. They did not exercise during the day prior to data collection.

2.2. Exoskeleton

The active lower-body K-SRD^TM exoskeleton straps to the user at the hips and legs (Fig. 1). The device utilizes sensors at the knee and hip joints to detect the user’s movement. Electric motors at the knee joints provide mechanical power to complement the user’s leg strength. Note that the K-SRD^TM does not initiate any movement. Rather, a feedback control algorithm processes the sensor measurements and computes appropriate control signals for the motors to output the necessary assistance for the user to complete the desired movement. For example, the exoskeleton will help push the user up after initiation of leg extension to stand, or will support the user after hip and knee flexion to sit. The K-SRD^TM weighs about 6.8 kg without the battery and about 9.1 kg with the battery pack needed to power the sensors and motors onboard the device. Battery life depends on intensity of the movement; approximately 6 hours during low-intensity activity and 1.5 hours during high-intensity activity. The commercially available and unobtrusive design of the K-SRD^TM exoskeleton provides advantageous lower-limb assistance for mobility to make it a viable candidate for use by workers.

Fitting of the exoskeleton involved adjustments for size and assistance. Modifications for hip width as well as thigh and shank lengths to assure maximal comfort and proper joint alignment preceded those for assistance. Adjustments include the amount of torque assistance provided for eccentric and concentric contraction during weight bearing. Default settings provide the greatest assistance during knee extension throughout stance and equivalent assistance for flexion and extension of the knee during swing phases. Level of assistance was modified according to feedback from each individual during basic movements, including walking, stepping in place, squatting, and kneeling. Once determined, the assistance level was not further modified.

2.3. Procedures

After completing informed consent forms, demographics, and health screening forms, participants donned a heart rate monitor (Polar H7: Polar Electro Inc., Lake Success, NY) and rested quietly for 5 minutes. Resting heart rate was determined as the lowest 15 second average during this time. A series of lifting tests were then performed to determine maximum acceptable weight (MAW), up to 23 kg, according to standard protocol [16], used in subsequent tasks. Next, participants were fitted with the exoskeleton (if being used that day) and passive reflective markers for movement recordings were placed bilaterally on the lateral aspects of the shoulders, hips, knees, and ankles overlying the greater tubercle, greater trochanter, lateral epicondyles, and lateral malleoli, respectively, to define trunk, thigh, and shank segments on left and right sides of the body. Warm-up included 5 minutes of brisk walking prior to task performance.

Participants stood upright in a comfortable position prior to performing each of three work tasks. Kneeling, lifting, and climbing represent activities associated with back and/or lower limb WMSDs and used in worker rehabilitation testing [17]. (1) Kneel task: step forward and kneel on one knee, holding the position for 5 seconds, then stand up before repeating the process 3 times on each leg. (2) Lift task: lift a box at MAW from floor to a waist level shelf back to floor 4 times within 20 seconds; (3) Climb task: step on and off a 20.3 cm (8-inch) stair (using an up, up, down, down method) for 15 cycles at a prescribed pace of one step per second, while holding a box at MAW. Although stepping on and off a single stair is not common in the workplace, the climb task does represent a work re-entry task post injury [17], hence a practical task to include. Due to the exploratory nature of this study, only basic instructions of task performances were provided to the participants. Tasks were performed in a random order. To ensure that heart rate returned to resting levels before starting each trial, time between trials and tasks varied by participant. Two participants completed these tasks with exoskeleton assistance (EXO) on the first day and without it (NONE) on the second day, while one completed the tasks without exoskeleton assistance on the first day and with it on the second day. Time of data collection over two consecutive days remained consistent within each participant.

2.4. Data collection

Data collection during performance included motion capture of movement kinematics at 100 Hz using a 4-camera Qualisys system (Qualisys Medical AB, Göteborg, Sweden) and heart rate data using the Polar H7 heart rate monitor. Data collection immediately after performance of each task included heart rate (HR) and ratings of perceived exertion (RPE) using the Borg-CR10 scale [18], the latter of which is deemed highly reliable in predicting worker fatigue [19]. After completing all tasks for each session, participants rated overall fatigue using a visual-analog scale from “well-rested” to “extremely fatigued”. After the session using the exoskeleton, participants completed a reliable short usability survey (System Usability Scale: SUS, [20]) and provided written feedback on their experience using the device and opinions about its comfort and function.

2.5. Data analysis

Quantitative objective results included movement kinematics and HR data. Subjective results included observations made during playback of motion capture data, RPEs, SUS scores, and written feedback.

Kinematics provided during task performances offered insight into movement similarities between conditions. Marker position data were processed through a second-order low-pass Butterworth filter with a 6-Hz cutoff frequency, as used elsewhere for whole body movements [21], prior to calculating hip and knee joint angles during task performances. Left and right hip angles were defined as the angles between trunk and thigh segments on the left and right side of the body, respectively. Left and right knee angles were defined as angles between thigh and shank segments on the corresponding side. Hip and knee kinematics were plotted across time with marker position data for each task, visually scanned, and marked to identify movement coordination used. Viewing playbacks of motion capture data offered further insight into movement coordination to verify interpretation of kinematic profiles. Flexion corresponded to decreases in a given angle, while extension corresponded to increases in a given angle. The amount of change in angle between two time periods represent the extent of flexion or extension. These were averaged across repetitions within subjects when reported as means and further averaged across participants when reported as group means. Step distance for the kneel task was determined as the difference between ankle location prior to movement initiation and after movement termination.

An investigator manually recorded HR, then an orally-reported RPE score immediately after each task performance. The SUS ratings followed accepted scoring methods to produce a percentile (0–100), where 68% is considered a general cut-off, such that scores higher than 68 are interpreted as above average usability [20].

3. Results

3.1. Movement kinematics

Movement variants across participants accompanied kinematic results for each task. Provision of basic instructions allowed for some different strategies to unfold, like those expected in the field.

3.1.1. Kneel task

Although all participants were told to step forward, kneel for 5 s, and then stand, two participants stepped back to start position after kneeling (P1, P2) and one stepped forward and turned around after kneeling (P3) prior to stepping for the next kneel. Mean step distance with EXO ranged from 16 cm less than to 6 cm greater than NONE and was not based on the forward or backward direction of the standing component (Fig. 2). P1 completed each step to kneel, hold, and stand in less time with EXO (8.9 s/step) than NONE (9.7 s/step), while P2 and P3 completed the task in more time with EXO than NONE (P2:9.8 vs 7.8 s/step and P3:8.8 vs 8.0 s/step, respectively). Leg angle (Fig. 3) varied by step distance (Fig. 2) differently for participants. Flexing the support knee with EXO a lesser amount than NONE corresponded to a slightly longer step distance with EXO for P1. Flexing the stepping or support knee or hip with EXO a greater amount than NONE corresponded to a longer step distance with NONE for P2. Less hip flexion with the stepping leg and greater knee flexion with the support leg for the left step with EXO corresponded to a longer step distance than NONE for P3, while greater hip flexion with the stepping leg and greater knee flexion with the support leg for the right step with EXO corresponded to a shorter step distance than NONE for P3. Hip flexion counter-movements of the stepping side ranging from 0–11 degrees (0–0.192 radians) often preceded hip and knee extension associated with standing. Average counter-movements for EXO and NONE were within 3 degrees (0.052 radians) of each other, thus very similar between conditions, yet differed up to 7 degrees (0.122 radians) for left and right steps.

Fig. 2.

Step distance for kneel task. Mean step distance for left (filled bars) and right (open bars) steps of the kneel task are shown for each participant for NONE (gray) and EXO (black) conditions. Error bars represent 1 standard deviation.

Fig. 3.

Flexion angles for the kneel task. Mean flexion of the hip (left panels) and knee (right panels) during the kneel task are shown for left (upper panel) and right (lower panel) steps. Positive values represent flexion, while negative values represent extension for NONE (gray) and EXO (black) conditions and left (filled bars) and right (open bars) sides of the body. Error bars represent 1 standard deviation.

3.1.2. Lift task

Figure 4 shows hip and knee angles across time for each participant performing one of the four cycles of the lift task for NONE and EXO. Plots between vertical lines represent the angle profiles during the lowering and lifting components of the task. Note the quick and large change in hip angle during lowering (flexion) and lifting (extension). Knee flexion fluctuated more and often began at or after the onset of hip flexion when lifting to reach peak flexion about the same time as the hip, while knee extension completed prior to or at the termination of hip extension when lifting. Amount of knee and hip flexion also varied according to the technique used. For NONE, P1 and P2 used the stoop technique, which requires most flexion at the hips with little knee flexion (see upper left panel, Fig. 5), while P3 used the squat technique, which requires extreme flexion at the knees and hips (see lower left panel, Fig. 5). For EXO, P1 and P2 increased and P3 decreased the amount of knee flexion, while participants decreased the amount of hip flexion on average relative to NONE (e.g., lower panels, Fig. 5). However, the amount of flexion varied across cycles between 4–21 degrees (0.070–0.367 radians) and depended on the joint of interest and the participant. Knee and hip flexion amounts varied most for P1, yet revealed the greatest knee flexion and/or least hip flexion for EXO in all but once cycle. The greatest between condition differences for P2 existed for knee flexion, in which EXO exceeded NONE by at least 22 degrees (0.384 radians), while the greatest between condition differences for P3 existed for hip flexion, in which NONE exceeded EXO by at least 6 degrees (0.105 radians). In each case, this led to the use of more of a semi-squat, a technique between the stoop and squat techniques. Additionally, lateral trunk flexion in NONE equaled that of EXO (P3 = 1.8 degrees (0.031 radians)) or exceeded that of EXO, yet remained relatively small (NONE vs EXO: P1 = 5.2 vs – 0.6 degrees (0.091 vs – 0.010 radians); P2 = 4.3 vs 2.1 degrees (0.075 vs 0.037 radians)).

Fig. 4.

Hip and knee angles for the lift task. Hip and knee angles are shown for one lower and lift cycle for each participant in NONE (left panels) and EXO (right panels) conditions. The left vertical segment in each panel represents the onset of lowering and the right vertical segment represents the end of lifting.

Fig. 5.

Squatting movement for the lift task. Side view of marker plots and connecting segments of the left side of P2 and P3 for NONE (left) and EXO (right) conditions during the lowering portion of the lift task are shown. Segments represent the trunk (shoulder to hip), thigh (hip to knee), and leg (knee to ankle). Black points and segments represent start and end positions, while gray represents the movement during task performance. The end positions for P2 NONE, P3 NONE, and P2 and P3 EXO represent stoop, squat, and semi-squat techniques, respectively.

3.1.3. Climb task

Knee and hip angle coordination of stepping during the climb task performances for EXO mimicked those for NONE (Fig. 6). Participants either stepped in a right, left, right, left pattern (P1, P3) or a left, right, left, right pattern (P2). Although hip and knee angles varied within conditions and among participants, a few consistent differences between NONE and EXO existed across the three participants. Knee flexion and extension of the trail leg with EXO exceeded that for NONE when stepping up onto the stair (compare black dashed plots at black arrows in Fig. 6 and upper panels in Fig. 7), while flexion and extension of the lead leg knee with EXO exceeded that for NONE when stepping down onto the floor (compare gray dashed plots at gray arrows in Fig. 6 and lower panels in Fig. 7). Greater flexion/extension of the lead leg for stepping up and for the trail leg when stepping down with EXO also exceeded NONE for P1 and P2. These movements allowed for greater foot clearance of the stair when wearing the EXO.

Fig. 6.

Hip and knee angles for the climb task. Hip and knee angles are shown for one step cycle of the climb task for P1 in NONE (upper panel) and EXO (lower panel) conditions. After flexing the lead leg joints to place the foot on the step (first vertical segment), participants transferred their weight to the lead leg and initiate extension just prior to initiating flexion of the joints of the trail leg to step up on the stair (black arrows, second vertical segment). Flexion of the lead leg joints to lift the foot (third vertical segment, gray arrows) preceded initiation of flexion of the trail leg joints to lower the body off the stair, then extension of the lead leg joints terminated to support the weight shift when stepping down to the floor (fourth vertical segment). The step down was completed when the trail leg extended for its placement on the floor at the end of the angle profile.

Fig. 7.

Knee angle change for the climb task. Mean changes in knee angle are shown for stepping up (upper panel) and stepping down (lower panel) of the climb task for NONE (gray) and EXO (black) conditions. Error bars represent 1 standard error. Square brackets indicate greater values for EXO than NONE for each participant.

3.2. Secondary outcomes

Table 1 shows the values for secondary outcomes of HR and RPE after task performances by each participant. Performing the kneel task resulted in small decreases in ending HR with EXO for two of three participants (–10 bpm relative to NONE) yet only P3 indicated a lower RPE for this task. Performance in the lift task produced no to moderate increases in HR with EXO. Interestingly, P2 with the moderate increase in HR (+11 bpm), was the only person to reveal a 0.5 decrease in RPE with EXO compared to NONE for the lift task. For the climb task, P2 reported a 0.5 higher RPE with a +20 bpm change in HR (EXO > NONE), while P1 and P3 reported similar RPEs for relatively small increases in HR with EXO. Apparently, RPE did not positively associate with small to moderate changes in HR in a meaningful way for these relatively short tasks.

Table 1
Heart Rate (HR) and Rating of Perceived Exertion (RPE) after task performance

Participant Kneel Lift Climb

NONE EXO NONE EXO NONE EXO

HR (RPE) 1 150 (0.5) 140 (1) 149 (1) 153 (1) 173 (2) 176 (2)

2 85 (0) 88 (0) 88 (0.5) 99 (0) 104 (0.5) 124 (1)

3 118 (3) 108 (2) 135 (3) 135 (3) 136 (4) 146 (4)

	Participant	Kneel	Lift	Climb
HR (RPE)	1	150 (0.5)	140 (1)	149 (1)	153 (1)	173 (2)	176 (2)
	2	85 (0)	88 (0)	88 (0.5)	99 (0)	104 (0.5)	124 (1)
	3	118 (3)	108 (2)	135 (3)	135 (3)	136 (4)	146 (4)

HR is in beats per minute. RPE was based on scores of Borg CR10 Scale [18].

Perceptions of fatigue and exoskeleton usability varied among participants. These data were collected at the end of each day of testing and followed performances of additional tasks, if time allotted. The two additional tasks for P1 and three additional tasks for P3 did not include kinematic assessments, thus were not reported here. P1 reported substantially lower fatigue using the exoskeleton (60.6% NONE vs. 37.1% EXO) while P2 reported essentially the same level of fatigue with and without the exoskeleton (8.96% NONE vs. 8.28% EXO). These participants rated usability of the exoskeleton as above average on the SUS (Total scores: P1 = 70 and P2 = 75, Table 2). However, P3 reported substantially higher fatigue using the exoskeleton (33.2% NONE vs. 73.7% EXO) and rated usability of the exoskeleton as below average (42.5, Table 2).

Table 2

Results of the System Usability Survey (SUS)

	Strongly Disagree	Disagree	Neutral	Agree	Strongly Agree
I think that I would like to use this system frequently.		P3	P1	P2
I found the system unnecessarily complex.	P1	P2	P3
I thought the system was easy to use.	P3			P1 P2
I think that I would need the support of a technical person to be able to use this system.		P2	P1		P3
I found the various functions in this system were well integrated.				P2 P3	P1
I thought there was too much inconsistency in this system.		P1	P2,P3
I would imagine that most people would learn to use this system very quickly.			P3	P1	P2
I found the system very cumbersome to use.			P1 P2 P3
I felt very confident using the system.			P1 P3	P2
I needed to learn a lot of things before I could get going with this system.	P2	P1 P3

4. Discussion

Results of this study address a gap in the literature by providing some evidence that people wearing a lower-body K-SRD^TM exoskeleton can complete common work tasks without major alterations in kinematics that may increase injury risk for the wearer. This project is the first step in establishing a future intervention for potentially mitigating back and lower extremity WMSDs in workers via the use of exoskeletons. The recent interest in industrial applications of exoskeletons ensues its successful applications for increasing the capabilities of ground soldiers [22] and for medical treatments such as movement-related improvements in spinal cord patients [23]. Exoskeletons provide an attractive alternative to current practices for performance of certain tasks since they are attached directly to users and do not require modifications to the work site, worker procedures, or conscious actions by the worker. With the expected emergence of HPA exoskeletons for tasks such as heavy lifting and manual labor in manufacturing, construction, warehouse, or other highly-constrained environments [11], comes the responsibility of ensuring safety for its users. Results of the current case study provide preliminary outcomes to introduce a much-needed protocol for evaluating the effectiveness of lower limb exoskeleton augmentation on kinematics when performing basic tasks associated with lower body WMSDs so common in industry.

Movement kinematics of hip and knee joints were quantified during performance of three fundamental tasks, with and without exoskeleton use. Results revealed that wearing the K-SRD^TM exoskeleton for a brief time period can, but does not have to, alter movement kinematics of the hips and knees during performances of the kneel, lift, and climb tasks. Although kinematic assessments on more participants are warranted, these initial outcomes indicate that the device can produce similar kinematics to performances without it. The discussion that follows uses the similarities and differences in movement kinematics obtained from the current study and results in the literature to offer no evidence of greater compromise in user safety with K-SRD^TM exoskeleton use for ordinary tasks.

One strategy to avoid injury when performing a lift task with a load involves avoiding lateral trunk flexion [24]. Compared to NONE, participants slightly decreased or matched lateral trunk flexion with EXO. Use of the EXO did not increase lateral trunk flexion kinematics to result in the use of this inappropriate, injury-producing movement in these participants. However, what do observations for altered kinematics indicate?

Results revealed similar directional likenesses in kinematics between EXO and NONE across participants for only one task. Two instances of greater knee flexion observed for the climb task occurred during the non-weight bearing phases of task performance, allowing for greater foot clearance. Carrying the MAW load during task performance blocked vision of the lower limb and likely contributed to increased foot clearance in both conditions initially, as observed during stair ascent with vision diverted away from the step [25]. Although the greater foot clearance observed with EXO compared to NONE may increase workload for participants, it also decreased the trip potential similar to young and older low risk fallers [26] and firefighters in unfatigued states [27].

The greater knee and/or hip flexion revealed with EXO for the kneel task corresponded often with step distance. Geometry dictates that greater knee and/or hip flexion accompanies kneel with the feet closer together. In this study, step distance represented the distance between start and end ankle location on the stepping foot, thus although step distance did not always represent the distance between feet, inverse relationships between hip or knee flexion and step distance were identified.

Alterations in relationships between hip and knee flexion for the lift task represented different lifting and lowering techniques (see results for explanation of stoop, squat, and semi-squat techniques) observed in the literature [28]. Although two participants used the stoop technique and one used the squat technique for NONE, techniques used by each person changed toward the semi-squat with EXO. The change from a squat to a semi-squat (P3) involving decreased knee flexion compared to changes observed for a similar lift task with increased weight [29]. Each lift technique presents with advantages and disadvantages. Large increases in hip flexion with little change in knee angles observed in the stoop technique can minimize energy expenditure [30] resulting in relatively low heart rates [31] during task performance. However, it may also produce high net moments, passive muscle or ligament forces, and compressive and shear forces on the spine [32], which could produce unwanted abnormal spinal movement and/or injury [33, 34]. In contrast, large increases in knee and hip flexion associated with the squat technique can accompany low net moments, passive muscle or ligament forces, and compression and shear forces on the spine [32]. Unfortunately, it can also be associated with greater energy expenditure [30] resulting in relatively high heart rates [31], thus hasten fatigue [35] during task performance. Changing to the semi-squat technique involves moderate levels of hip and knee flexion linked to decreased risk of muscle strains [31] and greater MAW [28]. Use of the semi-squat technique also produces heart rates less than those of the squat technique [31], lessens fatigue compared to the squat, and avoids extreme joint ranges of the squat and stoop [28]. Avoiding extreme flexion is considered by some advantageous for the lifting technique because it maintains erector spinae muscle activity in the back [36] which can protect spinal ligaments [37]. Apparently, appropriate training of the squat technique can diminish concerns for increased injury risk to passive tissues, which if performed as a quarter or half the range of a full squat, may increase such risk [38]. Although some researchers consider the semi-squat a good compromise between the stoop and squat [24, 28], technique use seems to be task dependent [31] and guided with appropriate lifting recommendations [24].

Participants experienced increased HR while wearing the EXO compared to NONE during the climb task. In addition, at the end of data collection, P2 stated that wearing the EXO made him “a bit hot”, while P1 and P3 stated that it seemed to take longer to reduce their HRs in the EXO. These outcomes correspond to the increases in HR, core temperature, and thermal sensation resulting from added protective body armor [39], which would likely intensify with greater activity. Support for this supposition comes from the slight decreases in HR with EXO for P1 and P3, who performed the kneel task early within a session before the squat and climb tasks, and the similar HR in EXO and NONE for P2, who performed the kneel task after the squat and climb tasks. Furthermore, with few exceptions, heart rate and perceived exertion did not differ substantially between EXO and NONE. While HR data limited to one endpoint reading per task do not estimate total energy expenditure, similarities between the two conditions suggest the K-SRD^TM exoskeleton may not increase metabolic demand for the selected tasks, despite the added weight of the device. Simpler devices supporting the ankle can reduce metabolic demand, while more complex devices supporting the knee and/or hip either do not change or increase metabolic cost [11]. The current findings likely result from the lighter and less bulky nature of the current device than earlier lower body exoskeletons.

Interestingly, the greater knee flexion to lift the leg in the climb task would normally activate exoskeleton assistance a few steps into the movement during swing for continuous forward stepping during walking or stair climbing. However, due to the continually changing testing conditions, which required up, up, down, down (forward, forward, backward, backward) stepping, it did not. Although this setup assists during tasks like walking, it may help explain the slightly greater HR observed for participants during performance of the climb task. The increased flexion combined with the increased mass of the exoskeleton requires greater work for the participant. Exoskeleton assistance during extension for stepping up and for flexion when stepping down (during weight bearing) was expected to counteract the weight of the device and decrease the overall workload during this task. Adjustments in settings and/or future alterations to EXO software for the climb task could allow users to benefit from EXO assistance during performance of this and similar tasks.

Researchers blame many WMSDs on fatigue. Negative relationships between fatigue and device usability exist such that greater usability associates with low levels of fatigue (e.g., [40]). For two participants, fatigue was rated low overall or decreased in EXO. These same two participants also rated usability of the exoskeleton as above average. However, the third participant rated fatigue to be higher with EXO and rated usability as below average, possibly due to the relative weight, fit, and/or device settings, as this person was smaller than P1 and P2 and mentioned some discomfort when sitting. Although preliminary, an inverse correlation between fatigue and value that workers place on exoskeleton practicality may exist.

5. Limitations

This study has several limitations. The number of participants in a case study always accompanies restrictions in the generalization of results. The outcomes of the present data only reveal that it is possible that a person wearing a lower-body K-SRD^TM exoskeleton can complete common work tasks without altering kinematics or with altering kinematics that do not appear to increase injury risk. Use of a lower limb exoskeleton does not provide direct support to the back. Although the objective of this case study was to address potential kinematic changes in EXO use, quantifying muscle activity and fatigue directly would allow for greater comparisons to other studies. Compared to a normal workday, participants had little time to become familiar with wearing the device. Over time, participants may change strategies for moving with the exoskeleton once accustomed to the type of support provided. This could further change kinematics and heart rate towards higher efficiency. Participants tested were all college students that do not perform full-time manual labor jobs. Although the climb task used in the study replicates a standardized task for worker rehabilitation assessment [17], it does not simulate continuous stair ascent (or decent) tasks more common in the workplace or allow the K-SRD^TM time to optimize forward or backward movements. Furthermore, the performance of tests over short periods of time do not reflect a typical workday for manual laborers. Extending the length of testing and recruiting numerous participants fully trained for full-time manual labor jobs would help evaluate the ability of the exoskeleton to alter movement kinematics and/or reduce physical demands in the appropriate population.

6. Conclusions

Performing common work tasks with an exoskeleton resulted in some alterations to kinematics while largely keeping heart rate and perceived exertion constant, implying that metabolic demands may remain similar to work performed without exoskeletons. This encourages progress in exoskeleton development, given that older, bulkier models can increase physical demands of the user, and the potential for exoskeleton use in the workplace. Additionally, two of the three participants rated their exoskeleton experience as having less fatigue and above average usability. This case study provides a basic framework for laboratory testing of exoskeletons used for industrial tasks. The results provide evidence of the need to continue developing and testing exoskeletons for possible applications into industrial settings to reduce physical demands while maintaining safe working conditions.

Conflict of interest

None to report.

Footnotes

Acknowledgments

The authors would like to thank Stephane Bedard and Nathaniel Zoso for their feedback regarding the K-SRD^TM activation and optimization and the LSU Office of Research and Economic Development, College of Engineering, and School of Kinesiology for financial assistance.

References

Center for Construction Research and Training (CPWR). The Construction Chart Book. 5th ed. Silver Spring, MD: CPWR-The Center for Construction Research and Training; 2013.

Boschman

, Frings-Dresen

MHW

, van der Molen

. Use of ergonomic measures related to musculoskeletal complaints among construction workers: A 2-year follow-up study. Saf Health Work. 2015;6(2):90–6.

Caban-Martinez

, Lowe

, Herrick

, Kenwood

, Gagne

, Becker

, et al. Construction workers working in musculoskeletal pain and engaging in leisure-time physical activity: Findings from a mixed-methods pilot study. Am J Ind Med. 2014;57(7):819–25.

Engholm

, Holmström

. Dose-response associations between musculoskeletal disorders and physical and psychosocial factors among construction workers. Scand J Work Environ Health. 2005;31(suppl 2):57–67.

Kramer

, Bigelow

, Carlan

, Wells

, Garritano

, Vi

, et al. Searching for needles in a haystack: Identifying innovations to prevent MSDs in the construction sector. Appl Ergon. 2010;41:577–84.

Gagnon

. The efficacy of training for three manual handling strategies based on the observation of expert and novice workers. Clin Biomech. 2003;18:601–11.

Bonato

, Ebenbichler

, Roy

, Lehr

, Posch

, Kollmitzer

, et al. Muscle fatigue and fatigue-related biomechanical changes during a cyclic lifting task. Spine. 2003;28(16):1810–20.

van der Molen

, Sluiter

, Hulshof

CTJ

, Vink

, Frings-Dresen

MHW

. Effectiveness of measures and implementation strategies in reducing physical work demands due ot manual handling at work. Scand J Work Environ Health. 2005;31(suppl 2):75–87.

Dale

, Miller

, Gardner

, Hwang

C-T

, Evanoff

, Welch

. Observed use of voluntary controls to reduce physical exposures among sheet metal workers of the mechanical trade. Appl Ergon. 2016;52:69–76.

10.

Albers

, Estill

, MacDonald

. Identification of ergonomics interventions used to reduce musculoskeletal loading for building installation tasks. Appl Ergon. 2005;36:427–39.

11.

Young

, Ferris

. State-of-the-art and future directions for lower limb robotic exoskeletons. IEEE Trans Neural Syst Rehabil Eng. 2017;25(2):171–82.

12.

Marinov

22 exoskeletons for work and industry into 6 categories Exoskeleton Report. 2016. Available from: http://exoskeletonreport.com/2016/04/exoskeletons-for-industry-and-work/.

13.

de Looze

, Bosch

, Krause

, Stadler

, O’Sullivan

. Exoskeletons for industrial application and their potential effects on physical work load. Ergonomics. 2016;59(5):671–81.

14.

Frost

, Abdoli

, Stevenson

. PLAD (personal lift assistive device) stiffness affects the lumbar flexion/extension moment and the posterior chain EMG during symmetrical lifting tasks. J Electromyogr Kinesiol. 2009;19(6):e403–12.

15.

National Institute of Standards and Technology (NIST). Robotic exoskeletons: The standards gap. 2016. Available from: https://www.nist.gov/news-events/news/2016/09/robotic-exoskeletons-standards-gap-0.

16.

Mayer

, Barnes

, Kishino

, Nichols

, Gatchel

, Mayer

, et al. Progressive isoinertial Lifting Evaluation I: A standardized protocol and normative database. Spine. 1988;13(9):993–7.

17.

Matheson

, Mooney

, Grant

, Leggett

, Kenny

. Standardized evaluation of work capacity. J Back Musculoskelet Rehabil. 1996;6:249–64.

18.

Borg

GAV

. Psychophysical bases of perceived exertion. Med Sci Sport Exer. 1982;14(5):377–81.

19.

Binoosh

, Mohan

, Bijulal

. Assessment and prediction of industrial workers’ fatigue in an overhead assembly job. S Afr J Ind Eng. 2017;28(1):164–75.

20.

Brooke

SUS: A ‘quick and dirty’ usability scale. In: Jordan

, Thomas

, McClelland

, Weerdmeester

, editors. Usability Evaluation in Industry: CRC Press; 1996. pp. 18994.

21.

Hondzinski

, Soebbing

, French

, Winges

. Different damping responses explain vertical endpoint error differences between visual conditions. Exp Brain Res. 2016;234(6):1575–87.

22.

Garcia

, Sater

, Main

. Exoskeletons for Human Performance Augmentation (EHPA): A program summary. J Robotics Soc Jpn. 2002;20(8):822–6.

23.

Miller

, Zimmermann

, Herbert

. Clinical effectiveness and safety of powered exoskeleton-assisted walking in patients with spinal cord injury: Systematic review with meta-analysis. Med Devices (Auckl). 2016;9:455–66.

24.

Burgess-Limerick

, Squat, stoop, or something in between? Int J Ind Ergonom. 2003;31:143–8.

25.

Shinya

, Popescu

, Marchak

, Maraj

, Pearson

. Enhancing memory of stair height by the motor experience of stepping. Exp Brain Res. 2012;223(3):405–14.

26.

Zietz

, Johannsen

, Hollands

. Stepping characteristics and Centre of Mass control during stair descent: Effects of age, fall risk and visual factors. Gait Posture. 2011;34(2):279–84.

27.

Kesler

, Horn

, Rosengren

, Hsiao-Wecksler

. Analysis of foot clearance in firefighters during ascent and descent of stairs. Appl Ergon. 2016;52:18–23.

28.

Straker

. Evidence to support using squat, semi-squat and stoop techniques to lift low-lying objects. Int J Ind Ergonom. 2003;31:149–60.

29.

Melino

, James

, Snodgrass

. The effect of load in a floor-to-bench lift during the WorkHab functional capacity evaluation. Work. 2014;49(4):585–96.

30.

Garg

, Herrin

. Stoop or squat - biomechanical and metabolic evaluation. AIIE Transactions. 1979;11(4):293–302.

31.

Wang

, Wu

, Sun

, He

, Wang

, Yang

. Squat, stoop, or semi-squat: A comparative experiment on lifting technique. J Huazhong Univ Sci Technolog Med Sci. 2012;32(4): 630–6.

32.

Bazrgari

, Shirazi-Adl

, Arjmand

. Analysis of squat and stoop dynamic liftings: Muscle forces and internal spinal loads. Eur Spine J. 2007;16(5):687–99.

33.

Skrzypiec

, Nagel

, Sellenschloh

, Klein

, Puschel

, Morlock

, et al. Failure of the human lumbar motion-segments resulting from anterior shear fatigue loading. Ind Health. 2016;54(4):308–14.

34.

Marras

, Lavender

, Leurgans

, Fathallah

, Ferguson

, Allread

, et al. Biomechanical risk factors for occupationally related low back disorders. Ergonomics. 1995;38(2):377–410.

35.

Kumar

. The physiological cost of three different methods of lifting in sagittal and lateral planes. Ergonomics. 1984;27(4):425–33.

36.

Mcgill

, Kippers

. Transfer of loads between lumbar tissues during the flexion-relaxation phenomenon. Spine. 1994;19(19):2190–6.

37.

Adams

, Dolan

. Recent advances in lumbar spinal mechanics and their clinical-significance. Clin Biomech. 1995;10(1):3–19.

38.

Hartmann

, Wirth

, Klusemann

. Analysis of the load on the knee joint and vertebral column with changes in squatting depth and weight load. Sports Med. 2013;43(10):993–1008.

39.

Larsen

, Netto

, Aisbett

. Task-specific effects of modular body armor. Mil Med. 2014;179(4):428–34.

40.

Rempel

, Star

, Barr

, Blanco

, Janowitz

. Field evaluation of a modified intervention for overhead drilling. J Occup Environ Hyg. 2010;7(4):194–202.