Design,development,and evaluation of a pneumatically actuated soft wearable robotic elbow exoskeleton for reducing muscle activity and perceived workload

Abstract

Introduction

To reduce muscle strain and enhance user comfort, a Pneumatically Actuated Soft Wearable Robotic Elbow Exoskeleton was developed. The design integrates human joint mechanics with a single-piece pneumatic structure for optimal assistance and reduced complexity.

Methods

Nineteen participants performed tasks under “Support On” and “Support Off” conditions to evaluate their effectiveness and feasibility. Muscle activity was measured using Surface Electromyography, while torque output and inflation times were assessed to gauge mechanical performance.

Results

The Soft Wearable Robotic Elbow Exoskeleton reduced muscle activity in the biceps by 22.36% and in the triceps by 18.19% at 18 PSI (Pounds per Square Inch). Torque tests revealed a maximum output of 4.39 Nm at 21 PSI. Inflation time tests showed that the exoskeleton could reach total inflation in as little as 0.22 seconds in high mode at 25 PSI, demonstrating the ability to achieve rapid actuation speed and efficiency.

Conclusions

These results indicate that the Soft Wearable Robotic Elbow Exoskeleton significantly reduces muscle activation, which may help mitigate the risk of work-related musculoskeletal disorders. By providing effective assistance during repetitive tasks, the exoskeleton enhances both mechanical performance and user comfort.

Keywords

pneumatic exoskeleton work-related musculoskeletal disorders (WMSD)electromyography (EMG)biomechanical testing/ analysis assistive technology human factors human subject trials

Introduction

Work-related Musculoskeletal Disorders (WMSDs) are caused by various factors, including manual material handling, awkward postures, repetitive motion, and overexertion, leading to conditions like pinched nerves, carpal or tarsal tunnel syndrome, and other connective tissue disorders.^1,2 Although days away from work (DAFW) due to WMSDs decreased from 2011 to 2018, likely due to increased ergonomic interventions and automation, WMSDs still accounted for 272,780 out of 900,380 DAFW cases in the U.S. private sector in 2018.² That shows 30% of all DAFW are related to MSDs.² In 2018, half of all private-sector WMSD cases were reported in retail trade, manufacturing, healthcare, and social assistance industries. The economic burden of WMSDs is significant, with annual costs in the United States estimated between $45 and $54 billion, including medical expenses, lost wages, and reduced productivity.^3,4 Workers suffering from WMSDs make nearly 70 million physician office visits annually, highlighting the healthcare demand these conditions create.³ One promising intervention to mitigate WMSDs is the use of exoskeletons.⁵

In the last 10 years, assistive Soft Wearable Robotic (SWR) devices have rapidly evolved, diversifying into areas like mobility assistance for daily activities, robot-assisted therapy, and human augmentation.⁶ In robotics, “soft” refers to materials that adapt to their environment and are flexible when subjected to external forces or interacting with objects of varying shapes.⁶ These wearable devices, which can be either active (powered) or passive (unpowered), are designed to augment and assist human movement, thereby reducing physical strain and the risk of injury.^5,7 Studies have shown that exoskeletons can significantly decrease muscle activity and perceived discomfort in tasks such as manual lifting and overhead work.⁸ Despite their potential, Exoskeleton adoption has been limited due to their cost, concerns about user comfort, and questions regarding their long-term applicability.⁹ Exoskeletons could significantly reduce WMSD-related injuries and associated costs. Large corporations like BMW, Ford, and Toyota have integrated approximately 585 passive exoskeletons to support their workforce, demonstrating productivity gains and reduced injury rates.¹⁰ However, broader adoption of exoskeletons, particularly active models, remains limited due to cost and operational challenges.¹⁰ Repetitive upper extremity movements are a well-documented risk factor for developing WMSDs, contributing significantly to muscle fatigue and strain in industrial tasks.¹¹ Current exoskeleton technologies have reduced physical demands, but many designs face challenges for widespread adoption. Rigid exoskeletons provide high force output but compromise comfort and portability, while soft exoskeletons offer comfort but may lack durability. Moreover, there is a shortage of lightweight, portable pneumatic exoskeletons for elbow-intensive tasks in industrial settings. To address these gaps, this study introduces a novel pneumatic elbow exoskeleton designed for repetitive lifting, assembly line work, and sustained arm support. Unlike existing devices, our design uses a single-piece pneumatic actuator with lightweight materials such as silicone and neoprene, optimizing both mechanical performance and user comfort. The exoskeleton reduces muscle activation, thereby mitigating strain and lowering the risk of WMSDs. This paper evaluates the benefits of our device, developed at The University of Texas at Arlington Research Institute, and aims to provide insights into the practical applications and benefits of active upper-body soft wearable robotic exoskeletons.

Related research

Research on exoskeletons has been ongoing for nearly 64 years since the 1960s, with significant progress in recent years.¹² While most of the research has focused on lower limb exoskeletons, limited progress has been made in upper limb exoskeletons due to the complexity of the human upper extremity.⁷ However, several key innovations have emerged to alleviate muscle strain through exoskeleton technology. The primary distinctions in exoskeletons are whether they are passive or active (unpowered or powered), rigid or soft, and their mode of actuation (fluidic or cable-driven).^6,8

Although little explored comparatively, Soft Wearable Robotic exoskeletons have been known to enhance human movement, augment physical capabilities, and support rehabilitation while being lightweight and versatile.⁶ Unlike traditional rigid exoskeletons, soft wearable robots utilize flexible, lightweight materials and innovative actuation mechanisms to provide more comfortable and adaptive assistance. Recent advancements in soft robotics have revolutionized the design and development of these wearable devices.¹³ High-force soft actuators are capable of generating significant force while maintaining a minimal weight profile, making them ideal for wearable robotics where comfort and minimal weight are crucial.¹⁴ The combination of soft and hard robotic elements in rehabilitation devices has shown improved motor function recovery for patients with movement impairments. This hybrid approach benefits from soft materials’ flexibility and hard components’ stability (Irshaidat et al., 2019).¹⁵

Nassour et al. (2021) investigated a fabric-based soft pneumatic elbow exoskeleton that significantly reduced muscle activity and metabolic cost during load-handling tasks. This suggests that such devices can improve worker efficiency and reduce fatigue in industrial environments.¹⁶ Research into pneumatic energy sources, such as compressed air tanks, has shown ways to power autonomous wearable exoskeletons effectively. This balances power supply and device portability, essential for practical use across various settings.¹⁷ Robotic exoskeletons have demonstrated efficacy in upper limb rehabilitation for stroke patients, improving muscle strength and coordination. This highlights the therapeutic potential of soft wearable robots in clinical environments (Rehmat et al., 2018).¹⁸ Thalman et al. (2018) developed a soft elbow exosuit made of TPU that enhances bicep lifting capacity, significantly reducing muscle strain during repetitive tasks. This highlights its potential in labor-intensive industries.¹⁹ Zhang et al. (2016) designed a soft wearable sleeve that improves joint mobility and reduces pain for patients with stiffness issues, broadening the application of soft wearable robots in managing chronic musculoskeletal conditions.²⁰ Zhao et al. (2022) introduced a novel interface based on soft skin sensors, enhancing the control and responsiveness of wearable exoskeletons. These innovations are crucial for improving user interaction and functionality.²¹ Despite these advancements, several challenges persist in soft wearable robotics. These include balancing portability, comfort, and efficiency and developing advanced control strategies for seamless human-robot integration. Future research should focus on refining actuator technologies, enhancing material properties, and improving sensory feedback systems to create more effective and user-friendly wearable robots.^13,14,16 The development of soft, wearable robotic exoskeletons marks a significant advancement in augmenting human capabilities and providing effective rehabilitation solutions. Ongoing innovation and interdisciplinary collaboration will be essential to overcoming current challenges and fully realizing the potential of these advanced assistive technologies.

In Industrial manufacturing settings, tasks such as repetitive lifting sustained overhead work, and prolonged static postures are major contributors to upper-limb Work-related Musculoskeletal Disorders (WMSDs).^22,23 These tasks place excessive strain on the elbow joint over time. Although pneumatic elbow exoskeletons have been explored, few usability studies have evaluated how effectively these devices fit real-world manufacturing needs or reduce WMSDs.²² Many existing designs employ multiple components and heavier materials, potentially limiting portability and comfort in repetitive, task-intensive environments. Our approach addresses these gaps by integrating a single-piece pneumatic actuator, lightweight neoprene, silicone materials, and a durable Onyx support plate to achieve controlled elbow bending. This combination enables simplified donning and improved comfort, allowing for prolonged use in settings where overhead and repetitive tasks are performed frequently. To benchmark our system against other soft wearable robotic elbow exoskeletons, we include a torque and usability comparison in a dedicated table, illustrating how our simplified design reduces complexity while delivering sufficient assistance. In this manner, our device targets specific manufacturing tasks that commonly lead to upper-limb injuries, pursuing the dual goals of mitigating injury risks and improving user comfort and efficiency. By focusing on portability, comfort, and minimal component requirements, our exoskeleton concept fills a critical niche among existing solutions, providing a platform for further development and validation in real-world industrial applications.

Materials and methods

Design and materials

Pneumatically Actuated Soft Elbow-exoskeleton (PASE) represents an advanced iteration of the soft elbow exoskeleton initially developed by Erel et al.²⁴ The design modifications focused on closely aligning the actuator with elbow biomechanics to enhance the integration with the limb and the effectiveness of the assistive motion. A unified actuator structure was adopted to minimize off-plane rotational forces, ensuring precise assistance with elbow movements and reducing joint strain. The actuator was constructed using XIAMETER® RTV-4234-T4 silicone from Dow Corning, chosen for its durability and flexibility. A base plate made of Onyx, a micro carbon fiber-filled nylon, supported the actuator, enabled controlled bending, and eliminated misalignment issues when the exoskeleton is worn.²⁵ A 1.5 mm thick contact layer of neoprene fabric featuring a soft, spongy microfiber core sandwiched between two smooth, tightly knit fabric layers was included to enhance wearability and user comfort, making it suitable for prolonged use.²⁶ The design improvements aimed at optimizing the exoskeleton’s performance and user experience, ensuring reliable and effective assistance during various tasks. Figures 1 and 2 illustrate the sectional and cross-sectional views along with the dimensions of the PASE (Table 1).

Figure 1.

(A) Isometric view of Actuator; (B) Isometric view of Base Plate; (C) Isometric view of Neoprene Layer.

Figure 2.

Isometric and Cross-sectional view of PASE.

Table 1.

Dimensions and Weight Components of PASE.

Components	Material	Length (mm)	Breadth (mm)	Depth (mm)	Weight (g)
Actuator	Silicone	270	79	78	420
Base plate	Onyx	310	95	2	46
Contact layer	Neoprene	310	95	2	106
Straps	Fabric	-	-	-	66
Total	-	-	-	-	638

Fabrication

The actuator’s fabrication process involved compression molding of XIAMETER® RTV-4234-T4 silicone. This design builds on the previous work of Erel et al. and Singh et al., where the fabrication process was conducted in two stages: the ridges were molded in the first stage, followed by the base in the second stage, allowing the creation of a cavity that enables uniform expansion of the actuator upon air inflation.^24,27 The base plate, made of Onyx, was produced using 3D printing technology. Neoprene fabric was cut to the dimensions of the base plate, with holes precisely aligned for assembly. The silicone actuator, Onyx base plate, and neoprene fabric were sewn to complete the actuator assembly (Figures 3 and 4). Finite element analysis (FEA) was employed to refine and optimize the design. This analytical approach comprehensively evaluated the actuator under various operational scenarios, guiding adjustments to enhance force delivery and movement support. These simulations identified and addressed potential design limitations, ensuring the actuator was finely tuned for optimal functionality.

Figure 3.

PASE assembly Configuration: (A) silicone actuator; (B) Onyx base plate; (C) neoprene fabric contact layer.

Figure 4.

Multifaceted views of the inflated soft elbow Exoskeleton: (A) Isometric view; (B) Side view; (C) Rear view.

Pneumatic control box

The exoskeleton’s operation is managed by a control box (Figure 5) designed to regulate airflow and pressure precisely. This control box includes two solenoid valves, a two-station manifold for directing airflow, and a proportional valve mounted on a single-station manifold for accurate pressure adjustments. A pressure sensor provides real-time monitoring, while a microprocessor handles data processing and command execution. The system uses an In-house compressed air, with the initial pressure regulated from 50 PSI to 25 PSI. The working pressure is then further controlled to 18 PSI using feedback from the pressure sensor. Once the desired pressure is achieved, the system autonomously stops the air inflow to maintain safety and ensure optimal support. The control system was isolated from the user and manually inflated during the test trials to focus on evaluating the mechanical assistance provided by the exoskeleton, with its weight not being considered in the evaluation.

Figure 5.

Pneumatic control system of PASE.

Mechanical testing

Torque test

An experimental setup (Figure 6) was developed to measure and characterize the torque generated by the exoskeleton. Torque measurements were taken at three angular positions (30, 60, and 90°) and three different pressures (15, 18, and 21 PSI). A load cell was mounted on the rotational axis of the exoskeleton, with a fixture at the endpoint applying perpendicular force to the load cell. A synthetic arm was created using a casting process incorporating 3D-printed bones and a revolute joint to replicate elbow movement. Alginate was used to cast the arm over the printed bones, forming a mold that produced the final silicone structure. The arm’s humeral segment (shoulder to elbow) measured 20 cm, and the forearm segment (elbow to fixture point) was 23 cm. The silicone arm weighs about 1.5 kg, which is low compared to a typical human arm. To minimize the effect of this weight, the wrist joint setup was mounted horizontally so that gravitational forces would not significantly influence the torque. Under these conditions, the arm’s mass did not substantially affect the torque data. We used a two-way ANOVA to examine the main effects of angle and pressure on torque, as well as their interaction.

Figure 6.

Experimental setup for Torque Evaluation.

Inflation time test

A similar setup measured the inflation time from 0 to 90° at varying pressures (20 and 25 PSI). The inflation speed was controlled by adjusting the electrical input to the solenoid, thereby regulating the airflow into the exoskeleton. Predefined modes (Low, Medium, and High) were established to vary the inflation speed. High mode corresponds to a fully open proportional valve, medium mode is half-open, and low mode restricts the valve opening to minimize airflow and inflation speed. This testing was crucial to accommodate different tasks and user needs, as not all scenarios require high-speed actuation.

Human subject testing

The human subject testing was aimed to evaluate the exoskeleton’s effectiveness in reducing muscle activation and perceived workload during various tasks. The study was approved by the University of Texas at Arlington Institutional Review Board (UTA IRB Protocol #:2024-0056), and informed consent was obtained from all participants.

Participant demographics

The study included 19 participants, 14 males, and 5 females, reflecting the demographic availability during recruitment. The participant’s ages ranged from 18 to 45 years, with the majority (16 participants) falling within the 18-25 age bracket. They represented diverse ethnic backgrounds, including Hispanic or Latino, Asian or Pacific Islander, and White. The participants had different physical activity levels, varying from minimal to over 10 hours per week. All participants were free from musculoskeletal disorders and had no health conditions that would restrict physical activity, such as lifting a 10 lb weight. A pre-screening process also confirmed that none of the participants had known skin allergies or allergies to the gel used in the study.

Experimental protocol

Data collection involved three distinct tasks, each lasting 70 seconds and structured into five repetitions of 8-second tasks interspersed with 5-second rest periods. Audio cues via a beep were provided to participants at each interval to indicate the start of an action, with a longer beep signaling the end of the activity. The initial repetition was discarded to minimize the influence of participants relying on muscle strength over the exoskeleton assistance. This setup allowed for comparing muscle activation and perceived workload between activities performed with and without the exoskeleton. Tests were performed with the device worn in unpowered and powered modes to examine practical usage scenarios of intermittent assistance. The natural case was not included because our primary focus was to compare how external support modifies muscle activity relative to an identical physical setup.

Electromyography (EMG) data were collected using the BIOPAC MP36 system during these tasks. Disposable circular Ag/AgCl electrodes, approximately 35 mm in diameter, were used for signal acquisition. Electrodes were placed according to SENIAM guidelines: biceps electrodes were positioned between the medial acromion and fossa cubiti. In contrast, triceps electrodes were placed at the midpoint between the posterior crista of the acromion and the olecranon.²⁸ Reference electrodes were placed near the wrist for consistency. Data was sampled at 1000 Hz and analyzed using MATLAB. In the data analysis, raw surface EMG signals were cleaned by discarding initial segments to remove transient artifacts, then passed through a 4th-order Butterworth bandpass filter (50–200 Hz) to reduce noise. The filtered signals were rectified, and any outliers (based on a Z-score threshold) were replaced with NaNs and linearly interpolated. Each participant’s peak muscle activation was subsequently identified, and a 4-second window before and after that peak was extracted to compare assisted (exoskeleton) and unassisted conditions. This peak-aligned approach provided a focused window of maximum muscle demand, enabling the calculation of percent reductions in EMG amplitude when exoskeleton assistance was applied.²⁹

Muscle activation testing for different tasks

The muscle activation testing involved three simulated occupational tasks designed to assess the exoskeleton’s impact on muscle activity under static and dynamic conditions. Static tasks involve controlled, isolated movements focusing on specific muscle groups with minimal variation in movement patterns. On the other hand, Dynamic tasks involve more complex, multidirectional movements that engage multiple muscle groups and require coordination and adaptability. Each task was tested under assisted conditions, where the exoskeleton provided support, and unassisted conditions, where participants performed the tasks without the exoskeleton’s aid. A constant pressure setting was used during task-based experiments, providing steady torque output for straightforward gravity support without active torque modulation. The order of assisted and unassisted trials was not randomized to maintain consistent conditions for each participant. Electromyography (EMG) was used to measure muscle activation in each scenario. The tasks included:

• TASK 1 - Weightlifting Task (Static Condition, Figure 7(A), and 7(B)): Participants lifted an 8-lb object from waist to chest height and held it for 8 seconds before lowering it. This task was designed to engage the biceps brachii and triceps brachii muscles in a controlled, targeted movement characteristic of a static condition.

• TASK 2 - Assembly Task with Basic Tools (Dynamic Condition, Figure 7(C), and 7(D)): Participants assembled three pre-drilled wooden pieces using nuts and bolts with a 2.6 lb hand tool, simulating a non-vibratory manual task. This task primarily engaged the forearm and shoulder muscles.

• TASK 3 - Assembly Task with Power Tools (Dynamic Condition, Figure 7(E), and 7(F)): Like the previous task, a 2.6 lb power tool introduces vibratory stress and assesses the exoskeleton’s ability to dampen vibration-induced muscle strain.

Figure 7.

Illustration of the evaluated tasks in the study: Task 1 - Weightlifting (A: Pre-lift, B: Post-lift); Task 2 - assembly with basic tools (C: Pre-lift, D: Post-lift); Task 3 - assembly with power tools (E: Pre-lift, F: Post-lift).

These tasks were selected to replicate typical manual material handling activities in manufacturing environments, focusing on repeated or sustained upper-limb movements. They were designed to evaluate the exoskeleton’s effectiveness in reducing muscle strain and perceived workload under realistic conditions. We calculated the relative difference in EMG amplitude (i.e., percentage change) for each participant during the “Support On” and “Support Off” conditions in the human subject tests.

Muscle activation comparison across different weights

A separate test was conducted to assess the exoskeleton’s performance further. A single participant lifted weights of 3, 5, 8, and 10 lbs. Based on preliminary testing, a pressure of 15 PSI was selected to ensure sufficient assistance and maintain user comfort. This test provided insights into how the exoskeleton handles varying loads and its effectiveness in reducing muscle activation across different weights.

Assessment of mental and perceived load

Two assessment tools were employed to evaluate the cognitive and physical workload and the user experience: the NASA Task Load Index (NASA-TLX) and a User Perception Survey.

NASA-TLX

The NASA-TLX survey was administered post-activity to gauge the participants’ perceived workload. This subjective workload assessment tool measures six dimensions: Mental Demand, Physical Demand, Temporal Demand, Performance, Effort, and Frustration. Participants rated each dimension on a scale from “Very Low” to “Very High”.³⁰ This comprehensive evaluation provided insights into the cognitive and physical demands placed on participants during the tasks with and without the exoskeleton. NASA-TLX dimensions are initially scored on a 0–20 range and later scaled to 0–100, where lower values imply reduced workload and higher values indicate greater perceived demand. The scores for each dimension were recorded and analyzed to understand the overall workload and identify areas where the exoskeleton may reduce strain. We will conduct pairwise t-tests to compare assisted and unassisted conditions across the three tasks.

User perception survey

After completing the tasks, participants were asked to complete a user perception survey. This survey assessed the exoskeleton’s utility, ease of use, comfort, and potential for extended use. The survey included the following items:

• Utility: Participants rated the exoskeleton’s usefulness in reducing physical demand on a scale from “Not at all Useful” to “Very Useful.”

• Ease of Use: Participants rated the exoskeleton’s ease of use from “Very Difficult to Use” to “Very Easy to Use.”

• Comfort: Participants rated the exoskeleton’s comfort level from “Not at all Comfortable” to “Very Comfortable.”

• Willingness to Use: Participants indicated their willingness to wear the exoskeleton for various durations, ranging from “Less than an hour” to “More than 5 hours a day.”

Additionally, an open-ended question allowed participants to provide comments or suggestions for improving the design of the exoskeleton to enhance safety and comfort. This qualitative feedback was crucial for understanding user experiences and identifying areas for future development.

Results

Torque test

The torque test results (Figure 8) demonstrate that the exoskeleton’s torque output increases with higher pressures (15, 18, and 21 PSI) and varies at angles of 30°, 60°, and 90°. At 15 PSI, the measured torque was 1.93 Nm, 1.53 Nm, and 1.10 Nm across the three angles. Increasing the pressure to 18 PSI yielded 3.15 Nm, 1.90 Nm, and 1.45 Nm, while at 21 PSI, the values rose to 4.39 Nm, 2.62 Nm, and 2.00 Nm. These observations confirm that higher pressures produce higher torque. To determine the significance of these differences, we used a linear regression model that included angle, PSI, and their interaction. This model accounted for 95.7% of the variance in torque ( $p = 0.0008$ ). PSI exhibited a strong positive effect ( $p = 0.0032$ ), and the angle × PSI interaction was also significant ( $p = 0.0329$ ), indicating that torque decreases with increasing angle, but to a degree that depends on the pressure level. A simpler model without the interaction explained 88.5% of the variance ( $p = 0.0015$ ), reinforcing that torque tends to decrease as the angle increases ( $p = 0.0024$ ) and increases as PSI rises ( $p = 0.0039$ ). These findings confirm that adjusting angle and pressure statistically impact torque generation, providing flexibility in meeting various task demands.

Figure 8.

Torque Output versus Angle for Different Pressures.

Inflation time test

The inflation time test results in Figure 9 demonstrate the time required to inflate the exoskeleton from 0 to 90° at different pressures and speed modes. At 20 PSI, the low mode took 5.26 seconds, the medium mode was 2.75 seconds, and the high mode was 0.24 seconds. When the pressure was increased to 25 PSI, the inflation times were slightly reduced to 4.99 seconds, 2.605 seconds, and 0.22 seconds for the low, medium, and high modes, respectively. These findings highlight the ability to achieve total inflation in as little as 0.22 seconds, indicating rapid actuation speed and efficiency.

Figure 9.

Grouped bar chart of inflation time by mode and pressure.

Muscle activation testing

The muscle activation testing results for the three tasks are presented in Table 2. The results indicate a consistent reduction in muscle activation for both biceps and triceps when using the exoskeleton. In Task 1, the average biceps and triceps activation reduction was 22.36% and 18.19%, respectively. Task 2 showed reductions of 14.41% for biceps and 11.37% for triceps, while Task 3 exhibited 10.21% and 5.22% reductions in biceps and triceps activation, respectively. These reductions demonstrate the exoskeleton’s efficacy in reducing muscle activity across various tasks. Figure 10 compares peak muscle activity for biceps and triceps across different tasks, comparing assisted and unassisted settings.

Table 2.

Muscle Activation Reduction Across Tasks for Biceps and Triceps.

Participant ID	Task 1		Task 2		Task 3
Participant ID	Biceps (%)	Triceps (%)	Biceps (%)	Triceps (%)	Biceps (%)	Triceps (%)
Participant 001	12.36%	15.40%	18.89%	16.48%	13.36%	21.52%
Participant 002	44.46%	44.30%	19.13%	28.99%	0.43%	0.31%
Participant 003	28.98%	24.09%	22.89%	16.99%	6.43%	−2.68%
Participant 004	37.35%	36.85%	24.13%	25.98%	12.35%	−6.81%
Participant 005	13.21%	8.68%	20.31%	21.51%	13.05%	4.49%
Participant 006	20.84%	16.60%	7.88%	9.92%	26.05%	3.07%
Participant 007	32.87%	31.36%	31.50%	32.33%	26.45%	−2.37%
Participant 008	20.62%	21.28%	11.74%	6.82%	22.87%	13.99%
Participant 009	14.61%	26.24%	21.74%	19.98%	7.95%	1.26%
Participant 010	2.05%	4.32%	1.71%	3.82%	1.12%	−7.08%
Participant 011	18.35%	8.55%	14.73%	1.88%	11.88%	8.47%
Participant 012	11.71%	21.07%	4.03%	9.68%	−0.23%	11.99%
Participant 013	16.16%	16.61%	14.93%	8.19%	27.94%	14.57%
Participant 014	25.73%	6.36%	7.86%	−0.38%	−16.78%	−5.59%
Participant 015	31.05%	14.13%	−3.20%	1.04%	20.23%	3.79%
Participant 017	23.74%	13.12%	16.51%	1.44%	−24.68%	27.24%
Participant 018	26.14%	8.90%	22.64%	6.46%	8.05%	11.33%
Participant 019	20.88%	13.30%	4.84%	4.74%	15.28%	−3.88%
Participant 020	23.70%	14.50%	11.55%	0.09%	22.17%	5.51%
Average reduction	22.36%	18.19%	14.41%	11.37%	10.21%	5.22%

Figure 10.

Task-wise peak muscle activity comparison - biceps and triceps.

Muscle activation comparison across different weights

The performance of the exoskeleton across varying weights was evaluated in a test conducted with a single male participant aged between 18 and 26 years. As shown in Table 3, the average reduction in muscle activation for the biceps was 24.18%, while the triceps showed a reduction of 19.24%. At 3 lbs, the reductions were 26.15% for the biceps and 21.60% for the triceps. At 5 lbs, the reductions were 26.62% for the biceps and 27.01% for the triceps. At 8 lbs, the reductions were 28.33% for the biceps and 14.49% for the triceps. At 10 lbs, the reductions were 15.64% for the biceps and 13.84% for the triceps. These results demonstrate that the exoskeleton effectively reduces muscle activation across various weights, although the degree of reduction varies with the weight.

Table 3.

Muscle Activation Reduction Across Different Weights.

Weight categories (Lbs.)	Muscle activation reduction
Weight categories (Lbs.)	Biceps (%)	Triceps (%)
3	26.15%	21.60%
5	26.62%	27.01%
8	28.33%	14.49%
10	15.64%	13.84%
Average reduction	24.18%	19.24%

Assessment of mental and perceived load

NASA-TLX

Participants completed the NASA-TLX in both “Support On” and “Support Off” conditions to assess the perceived workload. Across all tasks, average NASA-TLX scores were consistently lower with exoskeleton support. In Task 1, the assisted and unassisted means were $28.51 \pm 3.29$ and $37.89 \pm 4.26$ , respectively, resulting in a difference of $- 9.39 \pm 2.05$ at ( $t (18) = - 4.57, p = 0.0002$ ). In Task 2, assisted scores averaged $26.67 \pm 4.06$ compared to $35.53 \pm 4.89$ for unassisted, a difference of $- 8.86 \pm 2.43$ at $(t (18) = - 3.64, p = 0.0019) .$ For Task 3, the assisted mean was $22.32 \pm 3.11$ versus $31.93 \pm 4.65$ for unassisted, $- 9.61 \pm 2.62$ at $(t (18) = - 3.67, p = 0.0017)$ . Each difference reached statistical significance, indicating a clear reduction in mental, physical, and temporal workload when the exoskeleton was active. The consistently negative differences highlight that exoskeleton support significantly decreased NASA-TLX ratings across all tasks. These results emphasize the device’s potential to reduce perceived exertion and improve user comfort in repetitive or physically demanding activities (Figure 11).

Figure 11.

Aggregate NASA-TLX Workload Scores by Task (Standard Error) - Lower is better.

User perception survey

The user perception survey results evaluated the exoskeleton’s utility, ease of use, and comfort. In Task 1, the mean scores for usefulness, ease of use, and comfort were 5.17 (±0.23), 4.61 (±0.34), and 4.11 (±0.40), respectively. For Task 2, the scores were higher, with 5.89 (±0.24) for usefulness, 4.83 (±0.26) for ease of use, and 4.44 (±0.30) for comfort. Task 3 recorded mean scores of 5.28 (±0.33) for usefulness, 5.06 (±0.32) for ease of use, and 4.33 (±0.35) for comfort (Figure 12). These results reflect a generally positive user experience, highlighting the exoskeleton’s perceived effectiveness and comfort across different tasks.

Figure 12.

Aggregate User Perception Scores by Task (Standard Error) - Higher is better.

These findings highlight the exoskeleton’s capacity to reduce muscle activation and perceived workload during various tasks and weights, underscoring its potential to mitigate work-related musculoskeletal strain.

Discussion

The results of this study offer a thorough evaluation of the soft pneumatic elbow exoskeleton’s ability to reduce muscle activation, decrease perceived workload, and enhance user comfort during various tasks. This section explores these findings, highlighting their implications for exoskeleton design and application in occupational settings.

Performance evaluation

Torque testing demonstrated that the exoskeleton generates substantial torque across different pressures and angular positions. At 21 PSI, the exoskeleton achieved a maximum torque of 4.38,552 Nm at 30°, indicating its capability to provide significant assistance during elbow flexion tasks, such as lifting and holding objects. The adjustable pressure feature enhances the exoskeleton’s versatility, allowing customization for specific task requirements. The inflation time test results further emphasize this adaptability, showing varying inflation times across low, medium, and high modes (from 5.26 seconds at low mode and 20 PSI to 0.22 seconds at high mode and 25 PSI). This flexibility is advantageous for tasks demanding rapid actuation, such as assembly line work or heavy items.

Comparison of performance against other exoskeletons

The torque performance of PASE was benchmarked against other pneumatic exoskeletons. Table 4 reveals that while PASE operates at a lower pressure of approximately 145 kPa, it achieves a competitive torque output of 4.38,552 Nm. Ang et al. reported the highest torque of 28 Nm at 200 kPa, and Thalman et al. achieved a similar torque of 27.6 Nm at 300 kPa. Nassour et al.'s exoskeleton produced 3.6 Nm at 122 kPa, while Erel et al.'s earlier design reached 7.6 Nm at 150 kPa. Despite operating at lower pressures, PASE’s torque output is productive, especially given the safety and comfort considerations. Using silicone in PASE offers durability and flexibility, contributing to its efficiency. The variations in torque output across these exoskeletons can be attributed to differences in material selection, actuator geometry, and tested pressure, highlighting the influence of these factors on performance outcomes. This comparison underscores the potential for further optimization of PASE’s design, particularly through actuator configuration and testing pressure adjustments, to enhance its torque performance.

Table 4.

Torque performance comparison against other soft exoskeletons.

Model	Actuation type	Material	Max tested pressure (kPa)	Obtained torque (Nm)
Ang et al.¹³	Pneumatic	Thermoplastic polyurethane	200	28
Nassour et al.¹⁵	Pneumatic	Thermoplastic polyurethane	122	3.6
Thalman et al.¹⁸	Pneumatic	Thermoplastic polyurethane	300	27.6
Erel et al.²¹	Pneumatic	Silicone	150	7.6
PASE	Pneumatic	Silicone	144.79	4.39

User benefits

Muscle activation tests indicated a significant reduction in muscle strain when using the exoskeleton. On average, biceps and triceps activation decreased by 22.36% and 18.19%, respectively, during Task 1. These reductions were consistent across various tasks and weights, showcasing the exoskeleton’s efficacy in mitigating muscle load. Such reductions prevent work-related musculoskeletal disorders (WMSDs) familiar with repetitive or forceful motion tasks. The decreased muscle activation also implies that users can perform tasks with less effort, potentially boosting productivity and reducing fatigue. This is particularly important for tasks requiring sustained muscle activity, as the exoskeleton helps maintain performance levels over extended periods.

Analysis of negative values in muscle activation testing (task 3)

During Task 3, some participants exhibited negative values in muscle activation reduction, which could be attributed to several factors. One possible reason is the tightness of the straps, which may have restricted natural arm movement and affected muscle performance. The insufficient vibration dampening of the exoskeleton may have led to heightened vibrations, disrupting EMG signals, and potentially creating an impression of increased muscle activation. Variations in fit and user compatibility could have also affected consistent support. Further investigation is needed to improve performance and user experience in high-vibration tasks.

Comfort and user experience

The NASA-TLX scores indicate a significant reduction in perceived workload across all tasks when using the exoskeleton. Assisted conditions consistently showed lower mean scores than unassisted conditions, with reductions ranging from approximately 7 to 9 points. This suggests that the exoskeleton reduces physical strain and lessens cognitive and temporal demands, making tasks less taxing for users. User perception scores further highlight the exoskeleton’s positive impact on user experience. High usefulness, ease of use, and comfort scores suggest that participants found the exoskeleton effective and user-friendly. Comfort scores, while slightly lower than those for usefulness and ease of use, still indicate a generally positive user experience. This is crucial for ensuring compliance and long-term use in real-world settings. Feedback from participants pointed out areas for improvement, such as enhancing fit and reducing weight. These insights are valuable for future design enhancements to increase user comfort and satisfaction. Ensuring the exoskeleton is comfortable for prolonged use is essential for practical application, as discomfort can deter consistent use.

Study limitations

Several limitations were noted in this study. The participant group consisted of young adults aged 18 to 26 and was primarily male, potentially limiting the generalizability of the results to a broader age range of worker populations and genders. In addition, the study only examined short-term effects, leaving long-term usage impacts unexplored. Moreover, all tasks were conducted in a controlled laboratory setting, which may not fully reflect real-world conditions, such as harsh and rough factory environments. Additional research is necessary to assess these findings in actual work environments. One other possible limitation that might arise from our study is the omission of MVC measurements. We employed a within-subject design, and each participant performed tasks with and without exoskeleton assistance, allowing direct comparisons of relative reductions in muscle activation. As each participant served as their own control, individual differences in muscle capacity did not significantly impact our primary outcome.

Implications for exoskeleton design

The findings from this study have significant implications for exoskeleton design and application. Adjusting torque and inflation speed provides flexibility, allowing the exoskeleton to be customized for different tasks and user preferences. This adaptability is a crucial advantage, enabling the exoskeleton to be used in various occupational settings. Reducing muscle activation and perceived workload highlights the exoskeleton’s potential to prevent WMSDs and improve occupational safety. Lowering physical and cognitive task demands can enhance worker productivity and well-being. Future research should optimize the exoskeleton’s design based on user feedback and expand its applicability to other joints and movements. Additionally, long-term studies are needed to evaluate the sustained benefits and challenges of using the exoskeleton in real-world settings.

Conclusion

This study demonstrates the effectiveness of a Pneumatically Actuated Soft Elbow Exoskeleton (PASE) in reducing muscle activity and perceived workload during various tasks. Integrating human joint mechanics with a single-piece pneumatic structure, the exoskeleton provides significant assistance in elbow flexion, which is crucial for lifting and holding the elbow at a flex position. Mechanical testing revealed that PASE can generate substantial torque, with a maximum output of 4.39 Nm at 21 PSI. The inflation time tests further highlighted its ability to achieve total inflation in as little as 0.22 seconds, indicating precise control over actuation speed and efficiency. These capabilities ensure the exoskeleton can adapt to various task requirements, enhancing its versatility and practical applicability. Human subject testing indicated consistent reductions in muscle activation, with average decreases of 22.36% in the biceps and 18.19% in the triceps at 18 PSI applied pressure to the actuator. The NASA-TLX and user perception surveys confirmed that the exoskeleton significantly lowers perceived workload, with high usefulness, ease of use, and comfort scores, underscoring its potential for long-term adoption in real-world settings. Despite these promising results, further research is needed to address the study’s limitations, including the short-term nature of the testing and the controlled laboratory environment. Future studies should focus on evaluating the long-term impacts of PASE in diverse occupational settings and optimizing its design based on user feedback to enhance comfort and usability. Further design refinements will target improved portability and integrated control strategies for a broader range of tasks. In conclusion, the PASE exoskeleton represents a significant advancement in wearable robotic technology, offering a viable solution for reducing work-related musculoskeletal disorders by providing effective assistance during repetitive tasks. The findings from this study contribute valuable insights into the practical applications and benefits of soft wearable robotic exoskeletons, paving the way for further innovation in this field.

Footnotes

Acknowledgements

We want to thank the Industrial, Manufacturing, and Systems Engineering Department at The University of Texas at Arlington and the Biomedical Technologies Division at The University of Texas at Arlington Research Institute for their support and assistance in this research. We also acknowledge the UTA IRB for their guidance and approval (The University of Texas at Arlington’s Federal wide Assurance (FWA) number is FWA00001762, expiring November 10, 2027).

ORCID iDs

Eshwara Prasad Sridhar

Veysel Erel

Muthu B. J. Wijesundara

Ethical Statement

Author contributions

Conceptualization: E.S., V.E., M.B.J.W., M.R.; Methodology: E.S., V.E., M.B.J.W., M.R.; Validation: E.S., V.E.; Data curation, E.S., V.E.; Writing - Original Draft Preparation, E.S.; Writing - Review and Editing, E.S., V.E., A.N., M.B.J.W., M.R.; Supervision, V.E., M.B.J.W., M.R.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the UTA Interdisciplinary Research Program, which enabled collaboration between the Industrial, Manufacturing, and Systems Engineering Department at the University of Texas at Arlington and the Biomedical Technologies Division at the University of Texas at Arlington Research Institute.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

References

BLS OSH Definitions : U.S. Bureau of Labor Statistics . Bureau of labor statistics. https://www.bls.gov/iif/definitions/occupational-safety-and-health-definitions.htm, (accessed 6 July 2024).

Occupational injuries and illnesses resulting in musculoskeletal disorders (MSDs) : U.S. Bureau of Labor Statistics. Bureau of Labor Statistics . https://www.bls.gov/iif/factsheets/msds.htm (accessed 2 July 2024).

Read ‘musculoskeletal disorders and the workplace: low back and upper extremities’ at NAP.edu. DOI: 10.17226/10032.

The cost of musculoskeletal disorders (MSDs) [Infographic]. https://ergo-plus.com/cost-of-musculoskeletal-disorders-infographic/ (accessed 6 July 2024).

Madinei

Alemi

Kim

, et al. Biomechanical evaluation of passive back-support exoskeletons in a precision manual assembly task: “Expected” effects on trunk muscle activity, perceived exertion, and task performance. Hum Factors 2020; 62: 441–457.

Thalman

Artemiadis

. A review of soft wearable robots that provide active assistance: trends, common actuation methods, fabrication, and applications. Wearable Technol 2020; 1: e3.

Exoskeleton Catalog Archives . Exoskeleton report. https://exoskeletonreport.com/product-category/exoskeleton-catalog/ (accessed 6 July 2024).

Andrade

Nathan-Roberts

. Occupational exoskeleton adoption and acceptance in construction and industrial work: a scoping review. Proc Hum Factors Ergon Soc Annu Meet 2022; 66: 1325–1329.

Gonsalves

Akanmu

Shojaei

, et al. Factors influencing the adoption of passive exoskeletons in the construction industry: industry perspectives. Int J Ind Ergon 2024; 100: 103549.

10.

Marinov

. Passive exoskeletons establish A foothold in automotive manufacturing. Forbes. https://www.forbes.com/sites/borislavmarinov/2019/05/15/passive-exoskeletons-establish-a-foothold-in-automotive-manufacturing/ (accessed 6 July 2024).

11.

Lop

NSB

Salleh

Zain

FMY

, et al. Ergonomic risk factors (ERF) and their association with musculoskeletal disorders (MSDs) among Malaysian construction trade workers: concreters. Int J Acad Res Bus Soc Sci 2019; 9: 1269–1282.

12.

Gopura

RARC

Bandara

DSV

Kiguchi

, et al. Developments in hardware systems of active upper-limb exoskeleton robots: a review. Rob Auton Syst 2016; 75: 203–220.

13.

Shi

Dong

Lin

, et al. Soft wearable robots: development status and technical challenges. Sensors 2022; 22: 7584.

14.

Ang

BWK

Yeow

C-H

. Design and modeling of a high force soft actuator for assisted elbow flexion. IEEE Robot Autom Lett 2020; 5: 3731–3736.

15.

Mae

Soufian

Elkurdi

, et al. Soft and hard robotics for movement rehabilitation, analysis and modelling. In: 2019 12th International Conference on Developments in eSystems Engineering (DeSE), Kazan, Russia, 7-10 October 2019, 964–969. October 2019.

16.

Nassour

Zhao

Grimmer

. Soft pneumatic elbow exoskeleton reduces the muscle activity, metabolic cost and fatigue during holding and carrying of loads. Sci Rep 2021; 11: 12556.

17.

Wehner

Tolley

Menguec

, et al. Pneumatic energy sources for autonomous and wearable soft robotics. Soft Robot 2014; 1: 263–274.

18.

Rehmat

Zuo

Meng

, et al. Upper limb rehabilitation using robotic exoskeleton systems: a systematic review. Int J Intell Robot Appl 2018; 2: 283–295.

19.

Thalman

Lam

Nguyen

, et al. A novel soft elbow exosuit to supplement bicep lifting capacity. In: 2018 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Madrid, Spain, 1-5 October 2018, pp. 6965–6971.

20.

Zhang

Singh

Krishnan

. A soft wearable sleeve for joint stiffness modulation. In: 2016 IEEE International Conference on Advanced Intelligent Mechatronics (AIM), Alberta, Canada, 12–15 July 2016, pp. 264–269.

21.

Zhao

Liu

, et al. A novel human-robot interface based on soft skin sensor designed for the upper-limb exoskeleton. Proc Inst Mech Eng Part C J Mech Eng Sci 2022; 236: 566–578.

22.

Segar

Rahman

MNA

. Musculoskeletal symptoms and ergonomic risk assessment among production operators at manufacturing industries: a review. IOP Conf Ser Mater Sci Eng 2019; 607: 012009.

23.

Muggleton

Allen

Chappell

. Hand and arm injuries associated with repetitive manual work in industry: a review of disorders, risk factors and preventive measures. Ergonomics (Lond) 1999; 42: 714–739.

24.

Erel

Singh

Lindsay

, et al. Development and characterization of a modular soft actuator enabled elbow exoskeleton for assistive movements, In: Proceedings of the ASME 2021 International Design Engineering Technical Conferences and Computers and Information in Engineering Conference. Volume 8A: 45th Mechanisms and Robotics Conference (MR). Virtual, Online, August 17–19 2021. DOI: 10.1115/DETC2021-71549.

25.

Onyx - composite 3D printing material. https://markforged.com/materials/plastics/onyx (accessed 10 June 2024).

26.

What is neoprene rubber and what are its applications?

Process Industry Forum . https://www.processindustryforum.com/article/what-is-neoprene-and-why-should-we-use-it (accessed 10 June 2024).

27.

Singh

Erel

, et al. Development of soft pneumatic actuator based wrist exoskeleton for assistive motion. In: 2023 IEEE/ASME International Conference on Advanced Intelligent Mechatronics (AIM), pp. 359–366.

28.

Welcome to SENIAM. https://seniam.org/ (accessed 12 August 2024).

29.

Sridhar

Erel

Wijesundara

, et al. Evaluating an elbow exoskeleton for reducing work-related musculoskeletal disorders risk: an interdisciplinary study. IISE Annu Conf Proc. 2024; 1–6. DOI: 10.21872/2024IISE_6483.

30.

Hart

. Nasa-task load Index (NASA-TLX); 20 Years later. Proc Hum Factors Ergon Soc Annu Meet 2006; 50: 904–908.