1. Introduction

2. Materials and Methods

3. Research and Pre-commercial Devices

Figure 2. All wearable exoskeleton prototypes for wrist assistance. Devices are grouped by stiffness criteria in: rigid (a-c), soft (d-p) and compliant (q-t). From from top left to bottom right it is possible to find: a) Portable Wrist Exoskeleton (PWE) designed by [47]; b) eWrist designed by [13,17]; c) wrist exoskeleton designed by [18]; d) exosuit designed by [11] and already reported in [15]; e) Exo-Wrist designed by [19]; f) soft orthosis designed by [20]; g) Active Support Splint for the wrist (ASSIST) designed by [21]; h) Soft Origami-patterned Actuator (SOA)-based wrist exoskeleton designed by [22]; i) a soft wrist exoskeleton designed by [23]; l) a pneumatic soft wrist exoskeleton (EXOWRIST) designed by [37]; m) a soft robotic device designed by [34]; n) Soft Wrist Assist (SWA) designed by [24,25]; o) Wearable wrist and forearm exoskeleton developed by [26]; p) Advanced Service Robot (ASR) a Shape Memory Alloy (SMA)-based hand exoskeleton designed by [38]; q) SMA-based wrist exoskeleton designed by [27]; r) SCRIPT Passive wirst Orthosis (SCRIPT-SPO) designed by [28] already reported in [15]; s) a hand and wrist exoskeleton designed by [14,29]; t) a low profile wrist exoskeleton designed by [30].

Figure 2. All wearable exoskeleton prototypes for wrist assistance. Devices are grouped by stiffness criteria in: rigid (a-c), soft (d-p) and compliant (q-t). From from top left to bottom right it is possible to find: a) Portable Wrist Exoskeleton (PWE) designed by [47]; b) eWrist designed by [13,17]; c) wrist exoskeleton designed by [18]; d) exosuit designed by [11] and already reported in [15]; e) Exo-Wrist designed by [19]; f) soft orthosis designed by [20]; g) Active Support Splint for the wrist (ASSIST) designed by [21]; h) Soft Origami-patterned Actuator (SOA)-based wrist exoskeleton designed by [22]; i) a soft wrist exoskeleton designed by [23]; l) a pneumatic soft wrist exoskeleton (EXOWRIST) designed by [37]; m) a soft robotic device designed by [34]; n) Soft Wrist Assist (SWA) designed by [24,25]; o) Wearable wrist and forearm exoskeleton developed by [26]; p) Advanced Service Robot (ASR) a Shape Memory Alloy (SMA)-based hand exoskeleton designed by [38]; q) SMA-based wrist exoskeleton designed by [27]; r) SCRIPT Passive wirst Orthosis (SCRIPT-SPO) designed by [28] already reported in [15]; s) a hand and wrist exoskeleton designed by [14,29]; t) a low profile wrist exoskeleton designed by [30].

3.1. Rigid devices

3.1.1. Development of a Portable Wrist Exoskeleton (PWE)

Xiao et al. in [16,47] investigated the design and control of a 2 DoF (flexion/extension and radial/ulnar deviation) portable, active and rigid wrist exoskeleton conceived for post-stroke rehabilitation at home, or used by workers to provide assistance in industrial settings.

Figure 4. An older version of the Portable Wrist Exoskeleton (PWE) [47].

Figure 4. An older version of the Portable Wrist Exoskeleton (PWE) [47].

The Portable Wrist Exoskeleton (PWE) has a rigid kinematic chain made of ABS plastic and weighs 360 g. To avoid harm to the user, foam pads are inserted between the rigid parts and skin. To increase safety and prevent possible injuries, mechanical stoppers limit the RoM to $\pm 60$° flexion/extension, and $\pm 25$° radial/ulnar deviation. The power is transferred from the actuator (Pololu 298:1 micro DC geared motor), placed on the forearm brace, to the hand brace through gears and links. The maximum torque generated are 2.3 Nm for flexion/extension, and 2.5 Nm for ulnar/radial deviation, which are enough for rehabilitation application, yet it may not be adequate for industrial use.

Two different control modalities have been implemented: position control for passive repetitive pre-programmed movements, suitable for rehabilitation application; and a torque control designed for torque amplification applications such as for workers’ assistance. As control strategy, they investigated the feasibility of surface electromyography (sEMG) signal classification through neural network (NN) and support vector machine (SVM). The experiment consisted of measuring the sEMG amplitude and torque exerted by the main wrist muscles of a healthy subject, who gradually applied maximum voluntary contraction in all wrist directions with the forearm and hand placed on a rig. The SVM predicted and followed the wrist movement in real time with the greatest accuracy (up to 80.44%), exerting different levels of torque depending on the wrist position. This study suggests that machine learning techniques for motion prediction could be beneficial when performing highly dynamic tasks. However, the requirement for sEMG signals makes it hard to apply in industrial settings, where workers must wear them in contact with the skin.

3.1.2. The eWrist - A Wearable Wrist Exoskeleton

Lambelet et al. in [13,17] have presented eWrist, a portable and rigid 1-DoF wrist exoskeleton to support flexion/extension in rehabilitation and training. The device, shown in Figure 5, aims at enhancing wrist muscles activity in ADLs by measuring residual sEMG amplitude of stroke patients through a Myo armband: a commercially available sEMG device. The whole device consists of a kinematic chain made of rigid and soft 3D printed parts, and weighs 556 g, including the battery (80 g) and the Myo armband [17]. A Boa Closure is used as tightening system and allows quick and easy one-handed placement. The actuation system incorporates a 12V DC brushless rotary motor (Maxon EC 16) with a total reduction ratio of 475:1, making it difficult to back-drive. Therefore, mechanical transparency is obtained through active control. A two-channel Hall effect sensor, integrated within the motor, determines joint angular velocity and position. A load cell, mounted between the bevel gear and the hand fixation, determines the torque. The battery can power the device for about 125 min in normal use.

A real-time PD-controller, integrated in a Raspberry Pi Zero, implements an Assistance-As-Needed (AAN) support strategy. This adjusts position and torque output based on several inputs: raw sEMG data from the Myo, force from the load cell, and the angular velocity. Two different dynamic behaviours, for different rehabilitation settings, are available: transparent or resistive. In transparency mode users can move freely and rapidly with low interaction torques (up to 0.34 Nm); in resistive modality, movements are constrained by higher torque values (up to 1.59 Nm).

The device has been characterized based on standardized metrics for rehabilitation devices and, subsequently, tested with healthy and stroke participants. Fifteen healthy subjects (7 females and 8 males, mean age: 26 ±3.4) and two stroke survivors were recruited (both males, age: 68 and 52 years). Tests performed on healthy subjects showed that it can provide: a RoM of 154°, a maximum output torque of 3.7 Nm, a maximum output velocity close to 520°/s, and an average set-up time of 37.3 s. Observations from questionnaires showed that the eWrist was positively received: it helped impaired subjects achieve a RoM comparable to that of healthy subjects; all participants were able to use it independently; the fixation systems were evaluated as being efficient, secure, and easy to handle. However, some limitations still remain in terms of aesthetics, physical proportions and weight distribution: some subjects felt discomfort due to a wrong size, alignment mismatch between mechanical and biological joints, and skin pain caused by fixations.

3.1.3. Robotic Orthosis for Wrist Assistance

Sangha et al. in [18] have presented a wearable 1 DoF robotic orthosis to assist wrist flexion and extension in rehabilitation.

The device, shown in Figure 6, weighs 330 g and consists of a rigid aluminium kinematic chain, secured at the palm and forearm by C-shaped clamps and Velcro straps, and 3D printed ABS plastics which cover all the electronics. The actuator is a DC rotary motor with a custom gearbox with a high reduction ratio (1700:1). It provides a nominal torque of 1.12 Nm, and a stall torque of 8 Nm.

The device can operate in three different modes according to the patient’s impairment: passive, active resistive, and active assistive. Passive mode assists the wrist movements based on predefined parameters and the patient’s RoM. It is useful for those with muscle weaknesses. Active modes assist wrist movements to augment brain plasticity and decrease muscles spasticity. Assistance is provided after detecting muscular effort with 8 force sensitive resistors (FSR) to record force-myography (FMG) signals from the forearm. A neural network (NN) is implemented in Arduino software to process the FMG signals and send the control command to the motor.

Tested on a healthy volunteer, it has: a RoM from 0° (full flexion) to 120° (full extension), a nominal torque of 1.12 Nm, and a battery life of 150 min. Despite its interesting performances and features such as working modalities, compactness, lightweight, and cost, further investigations are required to test its efficacy in real scenarios with more subjects (both healthy and impaired).

3.2. Soft devices

3.2.1. Soft Wrist Exosuit

Chiaradia et al. [11] have developed a novel 1-DoF wrist exosuit for assistance at work. It is made up of a soft wrist orthosis and two 3D printed ABS supports (one for the back and one for the palm), which help to distribute pressure on large areas, bear cable tension and sensors. The wearable parts, shown in Figure 7, weigh 300 g.

The system is actuated remotely using a bowden cable transmission and a Kollmorgen motor (AKM23F) with an Apex Dynamics planetary gear drive (PG II 040), and a gear ratio of 10:1. This helps reduce the weight on the human joint and the metabolic impact. A load cell measures cable tension, and 9 DoF inertial measurement units (IMUs) estimate wrist flexion: one on the back of the hand and one on the forearm. The control strategy is based on admittance control for transparency and gravity compensation. The torque exerted on the wrist is compensated by evaluating an estimated torque, knowing the wrist angle and the cable force. The addition of a PID controller on the angular velocity gives the assistance faster by capturing the user motion intention.

The device was tested on four healthy participants (all males, aged 28.0 ±4.0) under both isometric and dynamic conditions. In the isometric test, the participants were asked to keep their forearm and hand parallel to the ground for 3 min holding 1.5 kg weight. While, dynamic test required participant to track a Minimum Jerk Trajectory (MJT) that involves a wrist flexion, from 0° to 60° at two different speeds (35 and 70°/s), with and without exosuit assistance, while performing 15 repetitions with a 1.5 kg of weight. Overall, the device enhances the wrist movement in a range of 150° (70° for flexion and 80° for extension) and can provide 3 Nm torque, sufficient to hold 3.0 kg. However, it adopts an external heavy DC motor which makes it difficult to wear, and the reduction in muscle activation seems lower than with other devices.

Despite its limitations, this prototype is worth attention due to its characteristics such as: industrial applications, softness because of the tendon-driven actuation, comfort and customization as the rigid parts are designed following a 3D scan of a human hand. Future developments may include full wearable actuation and control systems, improving the transmission efficiency with a control strategy for friction management and the comfort.

3.2.2. ExoWrist - a soft tendon-driven wrist wearable robot for dart-throwing motion

Choi et al. in [19] presented a new soft wearable wrist robot called ExoWrist. The device is active and conceived to restore wrist functionalities of weakened upper limbs after injuries by focusing on the Dart Throwing Motion (DTM), the most natural wrist movement [48,49,50,51]. It is expected to be used both in or out-of-clinics. The ExoWrist consists of a golf glove, a forearm active anchor and a wrist armlet. On the back of the glove, tendons are placed to transmit forces. The forearm anchoring consists of a truncated cone shape made of soft and flexible materials. This compresses the forearm only when assistance is needed to reduce force losses through friction. The wrist armlet is a silicon-based 3D printed part and is customisable based on: wrist width, wrist height, and the DTM orientation plane. The use of soft materials and cables makes the whole device comfortable with an overall weight of 1 kg. The DTM plane and wrist kinematics of each individual have been measured by using 3D motion-capture sensors (Vicon) to define the tendons path to correctly apply assistive force.

Figure 8. Soft tendon driven wearable wrist robot for dart-throwing motion designed and developed by [19]. Image adapted from [19].

Figure 8. Soft tendon driven wearable wrist robot for dart-throwing motion designed and developed by [19]. Image adapted from [19].

The robot has been assessed on three healthy subjects through three experiments, evaluating respectively: the efficacy of the anchoring system, the motion tracking performance, and the ability not to hinder other movements. Participants sat on chairs with their elbows extended, forearms pronated, and wrists in a neutral position. For the first experiment each participant was asked to wear the anchor and place their forearm on the setup, which restricts forearm movement. Then the maximum displacement of the forearm anchor was measured at different anchoring wire tensions. For the motion tracking control, the were instructed not to exert any volitional effort on their wrists, while the robot pulled them. The last experiment was the comparison of the average maximum ROM of the finger joints (by using a goniometer) and the grasp force before and after the actuation of the robot. The results showed that without the active anchor the robot could not provide proper assistance. When the anchoring point is activated and properly tightened, the robot can assist more than 0.5 Nm, otherwise it cannot generate more than 0.2 Nm, less than required for ADLs (0.35 Nm). The robot can extend the wrist along the DTM plane for more than 50°, more than is needed for constraint-induced movement therapy (CIMT, 0°-20°), and it does not also affect movements at elbow, shoulder and fingers.

3.2.3. A soft robotic orthosis for wrist rehabilitation

Bartlett et al. in [20] have proposed a home wrist rehabilitation soft device designed for patients suffering from hemiparesis after stroke. It is pneumatically actuated, portable, and soft, and weighs 2.26 kg. It consists of a glove, an elbow sleeve and a Boa ratchet tensioning mechanism, which facilitates donning and doffing with one hand and adaptation to different arm lengths. The actuation mechanism consists of pneumatic artificial muscles (McKibben actuators) anchored on both the palm and back of the hand. Their size is crucial: a tube diameter of 1/2 in. (approx. 12.5 mm) can exert a contractile force close to 120 N, at a pressure of 30 PSI (approx. 200 kPa), which is enough for rehabilitation. Their anchoring points determine the initial actuator length and thus affect contraction length, RoM and force direction.

Figure 9. Soft robotic wrist orthosis [20]. Image adapted from [20].

Figure 9. Soft robotic wrist orthosis [20]. Image adapted from [20].

The device works on agonist-antagonist principle: a single movement can be generated by activating a pair of actuators (e.g. the two in the palm for flexion, the two in the back for extension, etc.). The air pressure of each actuator is constantly monitored and modulated by a controller, which reads the status of pressure sensors and gives the input signal to a pump and relevant valves. The device is tested on a mannequin hand and on a healthy subject for assessment. The user was instructed to fully relax his wrist while the robot is actuated. Results show, it can support all wrist DoFs by providing assistance over a range of 91° in flexion/extension, 78° in pronation/supination, and 32° in radial/ulnar deviation. Presented to a group of stroke patients, the participants gave positive feedback for its use in therapy.

3.2.4. Active Support Splint driven by Pneumatic Soft Actuator (ASSIST)

Sasaki et al. [21] have developed ASSIST, an active soft wrist splint to assist elderly or people in need of care, making them more independent.

Figure 10. Outlook of the ASSIST device from [21]. Image adapted from [21].

Figure 10. Outlook of the ASSIST device from [21]. Image adapted from [21].

Two different types of ASSIST have been created: one for assistance in the whole RoM (type I), and the other for increasing muscular endurance (type II). They differ in the McKibben structure of artificial muscle actuators. Both devices consist of plastic interfaces, on the palm and back, to which two rotary soft pneumatic actuators are attached. Reinforcements at the ends allow the actuators to bend circumferentially, providing enough bending angle and torque for wrist assistance. The devices weigh almost 390 g. ASSIST is controlled by measuring the wrist angle with flex sensors and keeping the inner pressure constant. At 400 kPa pressure, type II provides almost 80° rotation and 1 Nm of torque; while type I allows the same bending angle with a lower torque (0.25 Nm). In contrast, the torque of type II decreases faster as the bending angle increases.

These devices were assessed on 5 male subjects by measuring. In these experiment, human wrist was bended by ASSIST without any human muscular effort. For evaluating the assist effectiveness, the bending angle at human wrist, and the amplitude of sEMG signals at the flexor carpi ulnaris were measured. Results showed their suitability in correctly bending the wrist, and reducing muscular effort while lifting 3 kg. Although the results are not statistically significant, these devices have promising capabilities also for industrial applications. However, there are drawbacks and further analysis and more data are needed to verify the benefits of prolonged use, make it fully wearable with a pressure tank and compressor above 400 kPa, assess reliability and control of the actuators’ behaviour.

3.2.5. A soft robotic wrist brace with origami actuators

Liu S. et al. in [22] designed a low-profile, active and soft robotic (SR) wrist brace, that is pneumatically driven and has 2 DoF (flexion/extension and radial/ulnar deviation). It consists of eight modular soft origami-patterned actuators (SOAs), a commercial wrist brace, and rigid anchors made of fabric to fix the actuators and transmit forces, as shown in Figure 11. Due to the inherent compliance of its materials, the SR brace enables safe interaction, has light-weight, compactness, comfort and adapts to various wrist sizes. The parts worn on the wrist weighs 214 g (each SOA weighs approximately 1.9 g), but the overall device weighs almost 1.76 kg, including the actuation system and batteries.

The actuation system includes four identical two-SOAs units and two diaphragm air pumps. The device works by alternately contracting and expanding the actuators (e.g. during flexion actuators on the dorsal side of the wrist elongate while those on the palmar contract). Their axial deformation exhibited when pressurized is converted into large bending due to anchoring constrain. The control architecture consists of a high-level and a low-level control. The high-level controller estimates the wrist position depending on the pressure feedback from each actuator, and compares it with the desired motion. Thus, a pressure command is sent to the low-level controller to regulate the SOA pressure and elongation.

The device assessment was based on the RoM, output force, wearing position adaptivity, and performance. The RoM was measured without and with external loads (100, 200, 300 g). The maximum values, at $\Delta P=160kPa$ pressure, were 30° in extension, 31° in flexion, 33° in radial deviation, and 22° in ulnar deviation. They decrease as the load increases. The force exerted achieved up to 7.5 N in flexion/extension, and 6 N in radial/ulnar deviation. The torque reached up to 0.76 Nm and met the functional requirement for rehabilitation therapy. The performance, assessed with IMU sensor, showed that the device consistently followed the planned flexion/extension, while it had less stability in ulnar/radial direction. Although the results are promising and comparable with existing devices, future work will aim to optimize the design, improve compactness and portability, validate the effectiveness and side effects of specific rehabilitation therapies.

3.2.6. Bioinspired Musculoskeletal Model-based Soft Wrist Exoskeleton

Ning Li et al. in [23] describe a novel soft wrist exoskeleton for stroke rehabilitation and ADLs, shown in Figure 12. They used commercially available components (motors, commercial body protectors, sensors, power supply) and investigated the distribution of muscle tension lines to identify the most efficacy path along which artificial tendons should be placed to correctly move the wrist. They analysed the kinematics of wrist muscles and simplified this to four main muscles: Extensor Carpi Radialis Longus (ECRL), Extensor Carpi Ulnaris (ECU), Flexor Carpi Radialis (FCR), and Flexor Carpi Ulnaris (FCU), which were arranged as to form a quadrilateral around the wrist. This design guarantees a more natural interaction with the user. To compare the exoskeleton movements with a real wrist, a VICON tracking system was used to capture wrist trajectories of healthy subjects and those produced by the exoskeleton mounted on a hand mannequin.

The assessment was done in simulation (OpenSim software), and on healthy and impaired subjects. One healthy volunteer and three stroke (hemorrhagic) survivors (3 males, age range 32-57 years) were enrolled for clinical trials. First of all, tests were done on a healthy subject as to have a baseline. The wrist motions were captured by VICON, and sEMG signals at forearm were measured when lifting 5 kg weight during wrist flexion with and without exoskeleton assistance. Then, tests on stroke patients were carried out to verify the clinical assistance effects. Overall, the device was able to cover the daily RoM requirements with a flexion/extension range of 115°, and a radial/ulnar deviation range of 70°. When tested on stroke patients, the device demonstrated an average 90.3% ability to recover healthy wrist motion in ADLs (e.g. drinking action was selected to evaluate the assistance effect of the exoskeleton). Moreover, by measuring the forearm sEMG signals from a healthy subject during 5 kg lift, the exo exhibited more than 40% reduction in muscle activation. The overall results are consistent both in simulation and real scenarios, and pave the way for new and even better performing soft wrist exoskeletons. Although it is designed for rehabilitation and daily life assistance, the halving of muscle effort in lifting 5 kg is a great achievement and similar solutions could also be adopted in industry.

3.2.7. EXOWRIST: a wrist exoskeleton actuated by pneumatic muscle actuators

Andrikopoulos et al. in [37] developed a novel soft 2-DoF robotic wrist exoskeleton for rehabilitation, powered by pneumatic muscle actuators (Mckibben actuators). The device, shown in Figure 13, consists of wearable elastic neoprene-based glove. It adopts four Pneumatic Muscle Actuators (PMAs), symmetrically distributed around the forearm, to function antagonistically and generate wrist flexion/extension and radial/ulnar deviation. PMA is like a tube, fixed to the glove with plastic supports, and it is characterized by a decrease in length when pressurized. The design uses few hard materials and enhances lightness, comfort and safety, with a total weight of approximately 430 g.

EXOWRIST’s performance have been evaluated on a healthy, passive male volunteer, with no medical history of hand and arm, to prove its motion capabilities and its safety. The motion test consists on the execution of wrist flexion-extension and ulnar-radial deviation movements, with the patient’s arm lying parallel to the floor. To achieve the maximum RoM, PMA should first be inflated to half its maximum stroke. Measured with an IMU placed on the back of the hand, the device can reach $\pm 30$° in both flexion/extension and radial/ulnar deviation in less than 1 s. The pressure range is 0 to 8 bar, with a maximum operating pressure reaching 630 N of delivered force. The control strategy adopts an Advanced Nonlinear PID (ANPID) algorithm which allows tracking of pre-defined sinusoidal motions with smooth, fast and accurate PMA responses. The safety was assessed by placing FSR force sensors between PMA connections and the human’s skin to measure the contact forces, as they are the main sources of torque. The shear forces generated remain low, not exceeding 2.2 N, and the respective generated torques remains under 0.055 Nm. The results prove that the EXOWRIST has potential in rehabilitation scenarios. However, there is a need for further improvements especially in the actuation system to make the device fully wearable, portable and safe, since high levels of pressure are required and must be supplied from a compressor or high pressure tank.

3.2.8. Carpal Tunnel Syndrome Soft Relief Device

Zhu et al. in [34] have proposed a novel device to alleviate Carpal Tunnel Syndrome (CTS) strains and pain by actively adjusting the wrist angle when operating in awkward postures for prolonged time, e.g. while typing on a keyboard. The device, shown in Figure 14, consists of an elastic fabric sleeve that can be worn like a glove, and two thermoplastic (TPU) airbag actuators (eight-flanged bladders) sewn onto it. They are located at the lower and top part of the sleeve to extend and flex the wrist respectively by dynamically pressurizing and depressurizing them.

The performances were assessed on a hand mannequin by lifting the hand, with an external load of 200 g, to a height similar to that of a keyboard (1.9 cm). The device was able to lift the hand above 2 cm under a pressure of 31 kPa and from 0° to around 65° in 9 s under a maximum pressure of 62 kPa. Although this device has interesting characteristics such as soft actuation, breathable materials, safety, easy and compact design, it needs further development to the design, control and experimental evaluation with subjects. CTS problems are highly topical and need special attention for the well-being of workers.

3.2.9. Wrist Assisting Soft Wearable Robot with integrated SMA Muscle

Jeong et al. in [24,25] proposed a novel shape memory alloy (SMA)-based wearable robot that assists 2-DoF (flexion/extension and radial/ulnar deviation) wrist motions in performing ADLs. SMA actuators are metallic alloy that deform when heated above their transformation temperature and reduce their length between 3-5%, depending on the type and shape of the alloy chosen. SMA has potential properties as artificial muscles since it can produce high forces and can be fairly rapidly actuated via Joule heating. Furthermore, if shaped as a coil spring, it can produce forces up to 10 N, a contraction ratio of 40% and strains over 200%, achieving performances higher that than of SMA shaped as wires [27].

The device, shown in Figure 15, named Soft Wrist Assist (SWA), consists of: a finger-less glove, a forearm Velcro strap, and an elbow anchoring Velcro strap to adapt to different users’ sizes, prevent and improve dislocation and slip [24]. Moreover, to transmit forces properly actuators are fixed on the glove with non-stretchable fabric. Five muscle-like actuators are attached at various positions: three to the back of the hand, and two on the palm. They are designed as coil springs, integrated into an active and stretchable coolant vessel, filled with mineral oil, for improved heating and cooling response. The wearable parts weigh 300 g, while the total mass, including the pump and radiator, is around 1.92 kg.

The device can produce combined wrist movements such as radial-extension and ulnar-flexion by selectively activating the actuators. First, the overall RoM, torque, mechanical performances, wearability and set-up time have been assessed on a hand mannequinn, and then on five healthy subjects (three males and two females). The ROM was measured through absolute magnetic encoders were attached to the wrist joint. The average RoM was 38°, 50°, 34°, and 35° respectively for flexion, extension, radial, and ulnar deviation. The maximum torques, measured on a 3D-printed arm mannequin, were 1.32 Nm during extension, while greater than 0.5 Nm for the other motions. Tests with external loads (1.5 kg and 3 kg) have shown that the torque assistance increased, on average, support by 62.81 %, 101.65 %, 58.11 %, and 44.23 % in flexion, extension, radial and ulnar deviation, respectively. The average wearing time was 87 s (if self-worn), and 75 s (if assisted by another person).

These performances are in line with rehabilitation targets. However,some issues have still to be solved: the anchoring system (Velcro straps) could not perfectly prevent dislocation and should be modified to ensure stronger fixation and faster locking. The robot size and shape should be optimized for all users, reducing discomfort due to actuator misalignments. Full wearability can be guaranteed by reducing weights and keeping working temperatures as low as possible to prevent burning of the user’s skin.

3.2.10. Wearable SMA-based Wrist and Forearm Exoskeleton

Hope and McDaid in [26] proposed a novel 3-DoF (flexion/extension, radial/ulnar deviation, pronation/supination) wearable and portable wrist-forearm SMA-actuated exoskeleton, for rehabilitation at home, or helping people perform ADLs. The exoskeleton is active, soft, compact, low profile, lightweight (950 g) and low noise. It is attached to the user at three points: hand, wrist and forearm. Forces/torques are transferred from SMA actuators to the limb by tendon-pulley mechanisms which adopt wheels of different radii for force amplification, arranged around the forearm. Each DoF is controlled independently through a tendon module and a SMA actuator, for a total of six tendons and related mechanisms. This arrangement allows complex combinations of movements (e.g. DTM) based on agonist-antagonist principles by varying SMA length through Joule heating. The wire tension is released after cooling through conduction and forced convection heat transfer with fans. Tendon length is chosen to allow free movements covering the maximum RoM, and a potentiometer, fitted into the amplification wheel, measures SMA wire linear displacement. Six force sensors, arranged around the hand, provide information on flexion/extension and radial/ulnar deviation. A variable stiffness model and active stiffness control of SMA actuators have been implemented. Stress and position of SMA wires are determined by potentiometer and force sensors. A closed loop PID controller modulates the duty cycle of the voltage applied to each actuator according to a target strain/position.

Figure 16. SMA wrist-forearm exoskeleton for home rehabilitation developed by [26]. Image adapted from [26].

Figure 16. SMA wrist-forearm exoskeleton for home rehabilitation developed by [26]. Image adapted from [26].

Two prototypes have been tested: one using a SMA actuator and a compensation spring, and another using SMA actuators in differential configuration. The assessment has been done by measuring tendon displacement on a test-bench while lifting a variable load and tracking different trajectories (step, sinusoidal and triangle waves). The expected displacement of the actuators in the spring-bias configuration covers 40 % of the expected RoM, while in the differential module covers the 65 %. In both cases the major issues are related to friction (especially at lower strain rates and duty cycle), tolerances in the mechanical components, and the uncontrolled pre-stress on the SMA wires. Further potential improvements could include: a quantitative analysis of system friction to generate better and smoother control; a more effective cooling system rather than miniature fans, integration of sEMG sensors on the forearm cover as additional sensing method; redesign of each module to reduce the overall weight and fit different sizes.

3.2.11. ASR: a wearable glove for hand grasping

Hadi et al. in [38] have presented the ASR (Advanced Service Robots), a 5-fingered SMA-based hand exoskeleton for grasping rehabilitation and assistance, for use both in clinics and at home. ASR is active, portable, light (300 g without electronics and batteries), compact, and noiseless. Its actuation system consists of: two fishing wires for each finger (attached to the proximal and distal phalanges), a SMA actuator for each fishing wire (for a total of 10 SMA), and guides for connecting wires and transferring the force. The SMAs are fastened to a rigid platform on the forearm, and use 1 m FLEXINOL of 0.38 mm diameter, that can produce 22.5 N force. When the actuators are heated up with a current of 2.2 A, their tension force and length variation are transformed into phalanges motion and hand grasping.

Figure 17. ASR: a SMA-based hand exoskeleton for grasping assistance [38]. Image adapted from [38].

Figure 17. ASR: a SMA-based hand exoskeleton for grasping assistance [38]. Image adapted from [38].

A theoretical model which correlates tendon tension and grasping force have been developed and experimentally assessed by using two load cells to measure fingertip and tendon forces, a signal amplifier and Arduino Uno micro-controller to record data. Results show good agreement between theoretical and experimental values. The device was assessed on a volunteer. The force exerted on the fingertip is 35% of the force produced by the SMA actuator. The total grasping force is more than 40 N, which is sufficient for typical ADLs (18 N). The overall speed of hand closure is 3 s, while it takes about 4 s to open by cooling down the actuators using air fans. Although not directly conceived for wrist assistance, due to motion synergies between hand and wrist, problems at the wrist level reduce grasping ability, therefore this device could be considered a valuable wrist support. Current drawbacks include: a lack of full wearability as integration of sensing and control systems is not yet implemented; no user trials; the high currents (2.2 A) needed which might be dangerous for real applications.

3.3. Compliant devices

3.3.1. SMA based wrist exoskeleton

Serrano et al. [27] have proposed a rehabilitation wearable wrist exoskeleton with 2-DoF (flexion/extension and radial/ulnar deviation) based on SMA actuators. The device, shown in Figure 18, is a hybrid because of a rigid kinematic chain around the joint actuated by flexible materials remotely placed. Flexinol^® is used as SMA actuator (0.51 mm of diameter), which can exert 35.6 N of force and more than 0.5 Nm of torque. One SMA wire is used for each movement, except flexion which is left under gravity. The exoskeleton is made of simple and low cost parts through 3D-sintering polyamide with aluminum powder. The rigid interfaces are sewn on a glove to ease wearability. The device is symmetrical and can be worn both on right and left hands. The overall weight, considering the actuators, is less than 1 kg and the price is approximatively 1060 $.

The feasibility of the system has been tested first on a simulator and then in real cases. Biomechanics of Bodies (BoB) software was used to select the proper actuation systems, evaluate human body biomechanics, mechanical designs and control algorithms. Then a pilot study was conducted on 3 healthy subjects. Results showed the device allows a RoM between + 40° and -10° in flexion/extension; while between + 30° and -10° in radial/ulnar deviation. In both cases, the exoskeleton can follow a reference movement with small error. For a proper displacement of the wrist, 2.2 m long SMA wires are needed for extension, and 1.7 m for radial/ulnar deviation.

The main difficulty when controlling SMA actuators is their hysteresis which introduces non-linearity in the system. Thus, a BPID controller (a combination of a standard linear PID controller with a bi-linear compensator) has been used for a single SMA wire. This device could be an alternative noiseless and low cost solution to current rehabilitation robots. Despite the inherent flexibility allows it to adapt easily to the body, the wearability is still an issue due to the encumbrance of long cables and the high temperature needed to activate the actuators.

3.3.2. SCRIPT: a passive orthosis

Ates et al. in [28] have developed a hand and wrist exoskeleton for post-stroke rehabilitation at home, which provides compliant and adaptable extension assistance during ADLs.

SCRIPT has been designed with either passive and active actuation. However, even if the active ones provide more benefits, their architecture results more complex, heavy (1.5 kg) and bulky. Thus, the authors focused on improving the passive mechanisms with dynamic interaction. The device in Figure 19 is called SPO-F, and represents the final achievement after 4 design architectures described in [28]. It is an hybrid solution involving a rigid kinematic chain with a soft actuation system (springs and cables). It provides assistance along 1-DoF: wrist and fingers extension to overcome the hyper-flexion problems and restore a more functional position. The finger mechanisms consist of 3D-printed stiff levers connected with digit caps via a Dyneema cable, and actuated via extension spring. The wrist mechanism is a 3D-printed double parallelogram which transfers torque to the hand plate thanks to an extension spring. Each spring force can be adjusted by individual ball-chains. As rigid interfaces, off-the-shelf ergonomic components from SaeboFlex [33] are used and available in different sizes (S, M, L, XL) to better fit on each subject.

The device RoM has been assessed by using rotary position sensors (potentiometers), an Arduino Nano micro-controller and a visual marker on the hand plate for motion tracking. Results show the device can rotate up to 45° in flexion and 30° in extension. Assistance is proportional to hand flexion, spring stiffness (k), levers length, and their placement. The forces and torques are measured, via force sensors, for different stiffness values and different pre-tensioning forces at fixed k = 0.5 N/mm. In all cases, the minimum torque is higher than 0.5 Nm, while the maximum is 2 Nm at 60° of extension.

The first SPO orthosis was tested by 33 stroke patients in 3 different EU-countries. This has helped address the final design of SPO-F, which looks lighter (650 g), safer, more professional, comfortable, compliant, simple, easy to wear and able to satisfy rehabilitation requirements according to a stroke patient. Despite the great achievements obtained over the years, the design should be further improved in compactness due to its vertical profile. Furthermore, the extension force applied on the digits should be assessed with more patients because the compression applied might cause some fingers pain.

3.3.3. Hand and Wrist actuated Exoskeleton for Rehabilitation and Training

Dragusanu et al. [14,29] have developed a 2-DoF (flexion/extension and radial/ulnar deviation) active and hybrid exoskeleton to allow people with disabilities regain autonomy.

The device, shown in Figure 20, consists of a tendon-actuated mechanism with thermoplastic interfaces, which allows remote actuation and user’s adaptation. All actuation and electronic components are placed on the forearm, and data are transmitted via Bluetooth. It is composed of two independent rigid parts tailored on the user: one on the hand and the other one on the forearm. Three tendons, wrapped around three pulleys, connect the motors on the forearm to the hand plate. Dynamixe XL-320 DC rotary motors are selected, with a stall torque up to 0.39 Nm at 7.4V, that is suitable for rehabilitation applications. The whole system weighs almost 300 g, and costs about 150 $. The control consists in tracking wrist movements by measuring the orientation of the hand with respect to the forearm using IMUs (on the hand and forearm). A Matlab GUI interface has been developed to guide the users during rehabilitation making the whole process easier to set and less boring.

This wrist exoskeleton has become a module of a full hand/wrist exoskeleton developed by the same authors [14], which can also actuate fingers flexion/extension, works for about 3 h, and has an overall weight of 500 g. The updated version of the wrist comes from users’ feedback on wearability and anatomical adaptability. Subsequently, all rigid interfaces and the actuation module have been re-designed with a parametric approach: pulleys connected to the motor shafts are reduced in dimensions; an external support is added to wrap the excess wire; an automatic closure for the forearm module is designed to ease and halve the wearing time. The structure and control are developed to guarantee the use of hand and wrist modulus independently.

The device’s performance has been assessed involving a patient in the execution of three exercises: flexion/extension, radial/ulnar deviation, and recording and reproducing a movement performed by a physiotherapist. Predefined set of movements are generated. Among 60 trials, the average root-mean-square (RMS) errors in following flexion/extension and radial/ulnar deviation movements are not normally distributed, and have statistically significant differences for different motor speeds ($\omega$). This device has interesting features and great potential for use in rehabilitation. It can assist all wrist movements, it is portable, wearable, cheap, lightweight, easy to control and manage autonomously, and has a TRL of 4. However, some improvements could be made to reduce the overall encumbrance, weight and improve the torque provided.

3.3.4. Low-Profile Two-DoF Wrist Exoskeleton

Higuma et al. in [30] have developed a 2-DoF rigid wrist exoskeleton for rehabilitation, which allows flexion/extension and radial/ulnar deviation movements.

Figure 21. Low profile 2-DoF wrist exoskeleton for rehabilitation developed by [30]. Image adapted from [30].

Figure 21. Low profile 2-DoF wrist exoskeleton for rehabilitation developed by [30]. Image adapted from [30].

The mechanism consists of a hand back support, a forearm base where two linear actuators are placed, and two steel spring blades which connect the motors to the hand. The device is inherently flexible thanks to elastic elements which can deform during motion and adapt to the wrist center of rotation while transmitting forces. Each linear actuator moves a spring blade independently, back and forth, allowing bi-directional force transmission.The device is made of 3D-printed resin, it is 310 mm in length and weighs 509.5 g.

Performances measured on a test bench showed a RoM in good agreement with Finite Element Analysis (FEA) results, which covers most of that of healthy subjects: 56.7° in flexion, 68.1° in extension, 39.5° in radial deviation and 13.8° in ulnar deviation. The constrains are mainly due to the limited stroke of the actuators. The torque varies from 0.26 Nm (max flexion) to 2.47 Nm (max extension), with an average of almost 0.65 Nm for radial and ulnar deviation. The maximum load applied perpendicular to the wrist is 10.24 N for flexion/extension, with a shear force of 7.98 N; while for radial and ulnar deviation is 4.26 N, with a shear force of 4.14 N. Despite small interaction forces, a human evaluation is required to verify whether it harms the user.

Overall, it is a simple mechanism with a fairly soft structure which allows the wrist moving less overloaded. However, since linear actuators are not manually backdrivable, when turned off the fixed position of the blades may apply some shear force on user’s skin. Moreover, the absence of a defined center of rotation could make the springs deform in unwanted manners, increasing discomfort.

4. Commercial Devices

4.2. Soft devices

4.2.1. Carbonhand^®

Carbon-Hand is an assistive soft robotic glove for use outside of clinical settings, built on the Soft Extra Muscle (SEM^TM) technology [35,69]. It is sold by Bioservo for almost $7000. It is designed on a glove with pressure sensors in the fingertips to measure contact forces when interacting with objects or tools. Thus, power will be applied when the user initiates gripping to ensure a firm grip. So, it activates 1 DoF: the gripping (or finger flexion). It augments human capabilities by applying a force of up to 20 N per finger (involving only 3 fingers). The overall device weighs 685 g (glove + control unit), but since the control unit and the battery could be placed wherever preferred to the user’s body, weight should not be an issue. Batteries are designed to last approximately 8 hours and the device is available in different sizes (XS, S, M, L, XL) for both right and left hands. The device is conceived for rehabilitation and assistance-at-home by helping people with reduced hand functions perform ADLs independently [35,69]. This device goes under our attention because, due to motion synergies, hand dexterity is related to wrist motion and resistance capabilities: problems at the wrist level reduce grasping ability. Therefore, this device could be considered a valuable wrist support.

Figure 27. Bioservo Carbonhand ^® device and examples of related application [54]. Legend: 1) SEM unit; 2) Cord; 3) Arm strap; 4) Slap wrap: 5) Glove, three finger version.

Figure 27. Bioservo Carbonhand ^® device and examples of related application [54]. Legend: 1) SEM unit; 2) Cord; 3) Arm strap; 4) Slap wrap: 5) Glove, three finger version.

A first pilot study in [70] involved 15 participants (18 - 65 years old) with a chronic SCI and impaired hand function. They were given instructions on how to use the device at home for 12 weeks for at least 4 h a day during regular ADLs, and executing task-specific activities such as squeezing and releasing a soft ball, simulated drinking, eating a meal and writing. Participants were asked to record their activities in a diary. They returned for reassessment after week 6 and week 12 to evaluate grip strength and hand function. Most participants reported that they used the glove 0.3–6 h daily, and the average grip strength across subjects improved from initial (9.9 ± 2.9 kg), to week 6 (14.0 ± 3.0 kg), and week 12 (14.0 ± 3.2 kg). Moreover, it has recently been assessed for six weeks in 63 participants (between 18 and 90 years old) with impaired hand functions [71]. The protocol consists of 5 assessments for each participant: 3 pre-assessments across three weeks as baselines prior to the intervention; a post-assessment Within 1 week of the end of the intervention; and a follow-up assessment 4 weeks later to measure the retention of effects. Participants are patients who experience limitations in hand function, and who will therefore be asked to use the glove, at least 180 minutes per week, during ADLs at home (such as lifting and carrying items, performing hobbies, cleaning, cooking). The outcome measures are handgrip strength, arm and hand functional abilities, amount of glove use, and quality of life. Preliminary results have shown promising improvements in grip strength (+27%), pinch strength (+15%) and hand functionality (+12%). Since 2022, it has also been approved as a medical device according to the European Medical Device Regulation (EU-MDR).

4.2.2. Ironhand^®

IronHand^® is a soft active exoskeleton for grasping assistance and augmentation also built on the Soft Extra Muscle (SEM^TM). Initially, the development of the SEM^TM technology was intended to rehabilitate patients with impaired hand function (e.g. CarbonHand^®). Today, Bioservo Technologies is also focusing on prevention of injuries at work. The product has undergone long-term testing with various industrial partners, which have become key factors for its quick development since its first release in 2019 for almost $6,500 [35,55]. IronHand 2.0, shown in Figure 28, consists of a sensorised glove, a back-pack (with control and battery) and a hip-carry, everything designed in different sizes (S, M, L, XL) to better fit user’s body. The whole system weighs almost 2,75 kg, of which 50 g the glove. The lithium battery in the back-pack is the sore point for the weight with a full-charge duration of almost 6-8 hours. The glove has force sensors (FSR) on the finger tips and the palm, and it is innervated by artificial tendons (e.g. bowden cables) which enhance fingers flexion and gripping thanks to the push-pull action of linear DC motors. Enabled when force sensors detect certain pressure levels, the tendon-driven system can generate a maximum force of 16 N per finger (80 N in total) [35,55], adjustable to adapt to different needs and applications, as shown in Figure 29. The device can collect and share data through Bluetooth, 4G and Wi-Fi among different devices (e.g. tablet, control equipment), and save them in a local storage or in cloud (BioCloud^TM), Figure 29. The collected data also allow to assess the wearer’s risk of developing injuries. Several clinical trials have been done on the SEM technology over months [54,55,70,71]. First preliminary study on the effect of IronHand in working tasks (automotive assembly) has been published in [72]. Eight participants (4 males and 4 females) were identified by a General Motors (GM) ergonomist based on the task which required the hand to be active for most of the work cycle, gripping efforts to manipulate parts/tools, and willingness to use the device for 2 weeks. Muscle activity was recorded using sEMG from main forearm muscles. For each participant, sEMG was collected on the first day without IronHand, and on the second day with it. Once all electrodes were properly placed, participants performed muscle specific isometric maximum voluntary contractions (MVCs) for 3 seconds. Controlling task, repetition and cycle variables across participants was not possible, since each one performed different tasks within a work cycle (almost 120 sec) on the assembly line for one 8-hour shift (e.g. curtain airbag installation and secure, floor-pan secure, overhead fastener secures, carry and installation of engine splash guard). Overall results have reported a significant reductions in forearm muscle activity, improvement in gross hand grip strength, pinch strength and all hand functions (e.g. grip, grasp, precise movements, writing, etc). Further evaluation demonstrated that 59% of the recorded cycles resulted in a reduction in at least one muscle’s activity and 41% in an increase in activity. Thus, when compared to no exoskeleton, the IronHand produced both increases and decreases in forearm muscle activity, depending on the individual and the specific tasks. Therefore, specific use cases need to be carefully determined and the device optimized for each individual to ensure benefits.

Even if it is a hand exoskeleton for grasping augmentation, again due to motion synergies, it is safe to assume that grip force and wrist fatigue are in a sense correlated. The strong the grasping the more compression will be transmitted to hand ligaments and wrist. However, in spite of these promising results, in our opinion, this does not mean that reducing grasping fatigue it will guarantee the same reduction in wrist workload. Rather, it will reduce the probability to get injured, inflammations (such as CTS) and pain in the wrist joint.

4.2.3. Paexo Wrist^®

Paexo Wrist is a commercial passive wrist exoskeleton developed and sold by Ottobock for almost 160$. In practice, it is an orthosis which aims at supporting the wrist while moving loads, and preventing injuries and inflammations. It can be used when holding a screwdriver, riveting tool or welding equipment, and carrying loads. By looking on the website Paexo.com and the user manual, the device adopts innovative solutions: materials for thermal regulation tested in space (provided by Outlast Material), a Pull-2-Lock mechanism for quick one-handed donning in few seconds, a metal splint inside the garment which fulfills the function of a flexible beam to absorb and transfer loads away from the wrist. For a better versatility, the device can fit both left and right hands and is available in different sizes (S, M, L) to satisfy different users. No sensors and control strategies have been implemented. The device and its application are shown in Figure 30. The company mentions that Paexo Wrist has been thought for relieving muscles and tendons when working for long periods with tools and in assembly, by stabilizing the wrist and ensuring an optimal distribution of the workload. Unfortunately we have found no more technical aspects or analyses about its positive effects on workers’ health. So that, we cannot quantify its usefulness compared to other devices.

5. Design proposal for a portable wrist exoskeleton

Figure 35. Control scheme articulated in two levels. The high-level is responsible for the assistance to be given within appropriate timing. The low-level determines which actuators to switch on and regulates the force/torque output references at the actuators.

Figure 35. Control scheme articulated in two levels. The high-level is responsible for the assistance to be given within appropriate timing. The low-level determines which actuators to switch on and regulates the force/torque output references at the actuators.

6. Discussion and conclusions

7. Appendix 1

Table 1. Comparison of all wearable and portable wrist exoskeletons highlighting their characteristics.

Table 2. Number of clinical trials, number of subjects involved, age, gender, setting and testing period for each device. The authors wish to emphasise that the information below may not be exhaustive, as it refers only to documentation accessible online, and does not include confidential data held by companies, especially for commercial devices.

Abbreviations

References

1. Gonçalves, H.; Cardoso, A.; Mattos, D.; Deola Borges, G.; Anacleto, P.; Colim, A.; Carneiro, P.; Arezes, P.M. Assessment of Work-Related Musculoskeletal Disorders by Observational Methods in Repetitive Tasks—A Systematic Review. Occupational and Environmental Safety and Health III 2022, 455–463. [Google Scholar] [CrossRef]
2. Ramazzini, B. De morbis artificum diatriba [diseases of workers]. American journal of public health 2001, 91, 1380–1382. [Google Scholar] [CrossRef]
3. Armstrong, T.J.; Fine, L.J.; Goldstein, S.A.; Lifshitz, Y.R.; Silverstein, B.A. Ergonomics considerations in hand and wrist tendinitis. The Journal of hand surgery 1987, 12, 830–837. [Google Scholar] [CrossRef]
4. Sluiter, J.K.; Rest, K.M.; Frings-Dresen, M.H. Criteria document for evaluating the work-relatedness of upper-extremity musculoskeletal disorders. Scandinavian journal of work, environment & health 2001, pp. 1–102. http://www.jstor.org/stable/40967163.
5. Franco, G.; Franco, F. Bernardino Ramazzini: The father of occupational medicine. American Journal of Public Health 2001, 91, 1382–1382. [Google Scholar] [CrossRef]
6. World Health Organization (WHO). Musculoskeletal conditions, Key facts. https://www.who.int/news-room/fact-sheets/detail/musculoskeletal-conditions, 2021.
7. INAIL-Italian Workers’ Compensatory Authority). OCCUPATIONAL EXOSKELETONS:WEARABLE ROBOTIC DEVICES TO PREVENT WORK RELATED MUSCULOSKELETAL DISORDERS IN THE WORKPLACE OF THE FUTURE, 2020. Available online:http://www.osha.europa.eu.
8. European Agency for Safety and Health at Work (EU-OSHA). Work-related musculoskeletal disorders – Facts and figures. Standard, European Agency for Safety and Health at Work (EU-OSHA), Publications Office of the European Union, Luxembourg, 2020. [CrossRef]
9. Pons, J.L. Wearable robots: biomechatronic exoskeletons; John Wiley & Sons, 2008; pp. 4,11,219,332. ISBN:9780470512944.
10. Du Plesis, T.; et al. Review of Active Hand Exoskeletons for Rehabilitation and Assistance. Robotics 2021, 10, 1–42. [Google Scholar] [CrossRef]
11. Chiaradia, D.; Tiseni, L.; Xiloyannis, M.; Solazzi, M.; Masia, L.; Frisoli, A. An assistive soft wrist exosuit for flexion movements with an ergonomic reinforced glove. Frontiers in Robotics and AI 2021, 7, 595862. [Google Scholar] [CrossRef]
12. Crea, S.; Beckerle, P.; De Looze, M.; De Pauw, K.; Grazi, L.; Kermavnar, T.; Masood, J.; O’Sullivan, L.W.; Pacifico, I.; Rodriguez-Guerrero, C.; et al. Occupational exoskeletons: A roadmap toward large-scale adoption. Methodology and challenges of bringing exoskeletons to workplaces. Wearable Technologies 2021, 2. [Google Scholar] [CrossRef]
13. Lambelet, C.; Temiraliuly, D.; Siegenthaler, M.; Wirth, M.; Woolley, D.G.; Lambercy, O.; Gassert, R.; Wenderoth, N. Characterization and wearability evaluation of a fully portable wrist exoskeleton for unsupervised training after stroke. Journal of NeuroEngineering and Rehabilitation 2020, 17, 1–16. [Google Scholar] [CrossRef]
14. Dragusanu, M.; Iqbal, M.Z.; Baldi, T.L.; Prattichizzo, D.; Malvezzi, M. Design, Development, and Control of a Hand/Wrist Exoskeleton for Rehabilitation and Training. IEEE Transactions on Robotics 2022. [Google Scholar] [CrossRef]
15. Pitzalis, R.F.; Park, D.; Caldwell, D.G.; Berselli, G.; Ortiz, J. State of the Art in Wearable Wrist Exoskeletons Part I: Background Needs and Design Requirements. Machines 2023, 11, 458. [Google Scholar] [CrossRef]
16. Xiao, Z.G.; Menon, C. Towards the development of a portable wrist exoskeleton. 2011 IEEE International Conference on Robotics and Biomimetics. IEEE, 2011, pp. 1884–1889. [CrossRef]
17. Lambelet, C.; Lyu, M.; Woolley, D.; Gassert, R.; Wenderoth, N. The eWrist—A wearable wrist exoskeleton with sEMG-based force control for stroke rehabilitation. 2017 International Conference on Rehabilitation Robotics (ICORR). IEEE, 2017, pp. 726–733. [CrossRef]
18. Sangha, S.; Elnady, A.M.; Menon, C. A compact robotic orthosis for wrist assistance. 2016 6th IEEE International Conference on Biomedical Robotics and Biomechatronics (BioRob). IEEE, 2016, pp. 1080–1085. [CrossRef]
19. Choi, H.; Kang, B.B.; Jung, B.K.; Cho, K.J. Exo-wrist: A soft tendon-driven wrist-wearable robot with active anchor for dart-throwing motion in hemiplegic patients. IEEE Robotics and Automation Letters 2019, 4, 4499–4506. [Google Scholar] [CrossRef]
20. Bartlett, N.W.; Lyau, V.; Raiford, W.A.; Holland, D.; Gafford, J.B.; Ellis, T.D.; Walsh, C.J. A soft robotic orthosis for wrist rehabilitation. Journal of Medical Devices 2015, 9. [Google Scholar] [CrossRef]
21. Sasaki, D.; Noritsugu, T.; Takaiwa, M. Development of active support splint driven by pneumatic soft actuator (ASSIST). Proceedings of the 2005 IEEE international conference on robotics and automation. IEEE, 2005, pp. 520–525. [CrossRef]
22. Liu, S.; Fang, Z.; Liu, J.; Tang, K.; Luo, J.; Yi, J.; Hu, X.; Wang, Z. A compact soft robotic wrist brace with origami actuators. Frontiers in Robotics and AI 2021, 34. [Google Scholar] [CrossRef]
23. Li, N.; Yang, T.; Yang, Y.; Yu, P.; Xue, X.; Zhao, X.; Song, G.; Elhajj, I.H.; Wang, W.; Xi, N.; et al. Bioinspired musculoskeletal model-based soft wrist exoskeleton for stroke rehabilitation. Journal of Bionic Engineering 2020, 17, 1163–1174. [Google Scholar] [CrossRef]
24. Jeong, J.; Yasir, I.B.; Han, J.; Park, C.H.; Bok, S.K.; Kyung, K.U. Design of shape memory alloy-based soft wearable robot for assisting wrist motion. Applied Sciences 2019, 9, 4025. [Google Scholar] [CrossRef]
25. Jeong, J.; Hyeon, K.; Han, J.; Park, C.H.; Ahn, S.Y.; Bok, S.K.; Kyung, K.U. Wrist assisting soft wearable robot with stretchable coolant vessel integrated SMA muscle. IEEE/ASME Transactions on Mechatronics 2021, 27, 1046–1058. [Google Scholar] [CrossRef]
26. Hope, J.; McDaid, A. Development of wearable wrist and forearm exoskeleton with shape memory alloy actuators. Journal of Intelligent & Robotic Systems 2017, 86, 397. [Google Scholar] [CrossRef]
27. Serrano, D.; Copaci, D.S.; Moreno, L.; Blanco, D. SMA based wrist exoskeleton for rehabilitation therapy. 2018 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). IEEE, 2018, pp. 2318–2323. [CrossRef]
28. Ates, S.; Haarman, C.J.; Stienen, A.H. SCRIPT passive orthosis: design of interactive hand and wrist exoskeleton for rehabilitation at home after stroke. Autonomous Robots 2017, 41, 711–723. [Google Scholar] [CrossRef]
29. Dragusanu, M.; Baldi, T.L.; Iqbal, Z.; Prattichizzo, D.; Malvezzi, M. Design, development, and control of a tendon-actuated exoskeleton for wrist rehabilitation and training. 2020 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2020, pp. 1749–1754. [CrossRef]
30. Higuma, T.; Kiguchi, K.; Arata, J. Low-profile two-degree-of-freedom wrist exoskeleton device using multiple spring blades. IEEE Robotics and Automation Letters 2017, 3, 305–311. [Google Scholar] [CrossRef]
31. JAS-Joint Active Systems. JAS Wrist. https://www.jointactivesystems.com/products/areas/wrist, 2017. Online; accessed 5th of August 2022.
32. Myomo Inc.. MyoPro Orthosis. https://myomo.com/what-is-a-myopro-orthosis/, 2006. Online; accessed 8th of August 2022.
33. Saebo Inc.. SaeboFlex. https://www.saebo.com/shop/saeboflex/. Last accessed 30th of November 2022.
34. Zhu, M.; Adams, W.; Polygerinos, P. Carpal tunnel syndrome soft relief device for typing applications. Frontiers in Biomedical Devices. American Society of Mechanical Engineers, 2017, Vol. 40672, p. V001T03A003. [CrossRef]
35. Bioservo Technologies. Bioservo Strength for life - Professional and Healthcare. https://www.bioservo.com/. Last accessed 7th of September 2022.
36. Ottobock. Paexo Wrist Exoskeleton. https://paexo.com/paexo-wrist/?lang=it. Last accessed 5th of August 2022.
37. Andrikopoulos, G.; Nikolakopoulos, G.; Manesis, S. Motion control of a novel robotic wrist exoskeleton via pneumatic muscle actuators. 2015 IEEE 20th Conference on Emerging Technologies & Factory Automation (ETFA). IEEE, 2015, pp. 1–8. [CrossRef]
38. Hadi, A.; Alipour, K.; Kazeminasab, S.; Elahinia, M. ASR glove: A wearable glove for hand assistance and rehabilitation using shape memory alloys. Journal of Intelligent Material Systems and Structures 2018, 29, 1575–1585. [Google Scholar] [CrossRef]
39. Ferguson, R.; et al. Wrist pain: a systematic review of prevalence and risk factors–what is the role of occupation and activity? BMC Musculoskeletal Disorders 2019, 20. [Google Scholar] [CrossRef]
40. Noronha, B.; Accoto, D. Exoskeletal devices for hand assistance and rehabilitation: A comprehensive analysis of state-of-the-art technologies. IEEE Transactions on Medical Robotics and Bionics 2021, 3, 525–538. [Google Scholar] [CrossRef]
41. Gopura, R.; Kiguchi, K. Mechanical designs of active upper-limb exoskeleton robots: State-of-the-art and design difficulties. 2009 IEEE International Conference on Rehabilitation Robotics. IEEE, 2009, pp. 178–187. [CrossRef]
42. Gopura, R.; Bandara, D.; Kiguchi, K.; Mann, G.K. Developments in hardware systems of active upper-limb exoskeleton robots: A review. Robotics and Autonomous Systems 2016, 75, 203–220. [Google Scholar] [CrossRef]
43. Nikhil, G.; Yedukondalu, G.; Rao, S. Robotic exoskeletons: a review on development. International Journal of Mechanical and Production Engineering Research and Development 2019, 9, 529–542, https://archive.org/details/52.ijmperdaug201952. [Google Scholar]
44. Tiboni, M.; Borboni, A.; Vérité, F.; Bregoli, C.; Amici, C. Sensors and actuation technologies in exoskeletons: A review. Sensors 2022, 22, 884. [Google Scholar] [CrossRef]
45. Hussain, S.; Jamwal, P.K.; Van Vliet, P.; Ghayesh, M.H. State-of-the-art robotic devices for wrist rehabilitation: Design and control aspects. IEEE Transactions on Human-Machine Systems 2020, 50, 361–372. [Google Scholar] [CrossRef]
46. Cornejo, J.; Huamanchahua, D.; Huamán-Vizconde, S.; Terrazas-Rodas, D.; Sierra-Huertas, J.; Janampa-Espinoza, A.; Gonzáles, J.; Cardona, M. Mechatronic exoskeleton systems for supporting the biomechanics of shoulder-elbow-wrist: An innovative review. 2021 IEEE International IOT, Electronics and Mechatronics Conference (IEMTRONICS). IEEE, 2021, pp. 1–9. [CrossRef]
47. Khokhar, Z.O.; Xiao, Z.G.; Menon, C. Surface EMG pattern recognition for real-time control of a wrist exoskeleton. Biomedical engineering online 2010, 9, 1–17. [Google Scholar] [CrossRef] [PubMed]
48. Vardakastani, V.; Bell, H.; Mee, S.; Brigstocke, G.; Kedgley, A.E. Clinical measurement of the dart throwing motion of the wrist: variability, accuracy and correction. Journal of Hand Surgery (European Volume) 2018, 43, 723–731. [Google Scholar]
49. Braidotti, F.; Atzei, A.; Fairplay, T. Dart-Splint: an innovative orthosis that can be integrated into a scapho-lunate and palmar midcarpal instability re-education protocol. Journal of Hand Therapy 2015, 28, 329–335. [Google Scholar] [CrossRef]
50. Kaufman-Cohen, Y.; Friedman, J.; Levanon, Y.; Jacobi, G.; Doron, N.; Portnoy, S. Wrist plane of motion and range during daily activities. The American Journal of Occupational Therapy 2018, 72, 7206205080p1–7206205080p10. [Google Scholar] [CrossRef]
51. Brigstocke, G.; Hearnden, A.; Holt, C.; Whatling, G. In-vivo confirmation of the use of the dart thrower’s motion during activities of daily living. Journal of Hand Surgery (European Volume) 2014, 39, 373–378. [Google Scholar]
52. Borislav, M. The Number Of Companies Making Industrial Exoskeletons Has Been Quietly Increasing For The Past Five Years. https://www.forbes.com/sites/borislavmarinov/2020/09/24/the-number-of-companies-making-industrial-exoskeletons-has-been-quietly-increasing-for-the-past-five-years/?sh=231b1c067bf4, 2020. Last accessed 5th of August 2022.
53. Borislav, M. Updated Directory of Exoskeleton Companies and Industry Statistics. https://exoskeletonreport.com/2021/12/updated-directory-of-exoskeleton-companies-and-industry-statistics/, 2021. Last accessed 5th of August 2022.
54. Radder, B.; Prange-Lasonder, G.B.; Kottink, A.I.; Holmberg, J.; Sletta, K.; van Dijk, M.; Meyer, T.; Melendez-Calderon, A.; Buurke, J.H.; Rietman, J.S. Home rehabilitation supported by a wearable soft-robotic device for improving hand function in older adults: A pilot randomized controlled trial. PloS one 2019, 14, e0220544. [Google Scholar] [CrossRef]
55. Hyttinge, M.; Einarsson, F. Bioservo Technologies. Technical report, Redeye, Equity Research and Investment Banking Company, 2021.
56. Moskiewicz, D.; Mraz, M.; Chamela-Bilińska, D. Botulinum Toxin and Dynamic Splint Restore Grasping Function after Stroke: A Case Report. International Journal of Environmental Research and Public Health 2023, 20, 4873. [Google Scholar] [CrossRef] [PubMed]
57. Lucado, A.M.; Li, Z.; Russell, G.B.; Papadonikolakis, A.; Ruch, D.S. Changes in impairment and function after static progressive splinting for stiffness after distal radius fracture. Journal of Hand Therapy 2008, 21, 319–325. [Google Scholar] [CrossRef]
58. Lucado, A.M.; Li, Z. Static progressive splinting to improve wrist stiffness after distal radius fracture: a prospective, case series study. Physiotherapy theory and practice 2009, 25, 297–309. [Google Scholar] [CrossRef] [PubMed]
59. Berner, S.H.; Willis, F.B. Dynamic splinting in wrist extension following distal radius fractures. Journal of orthopaedic surgery and research 2010, 5, 1–4. [Google Scholar] [CrossRef]
60. Schwartz, D.A. Static progressive orthoses for the upper extremity: a comprehensive literature review. Hand 2012, 7, 10–17. [Google Scholar] [CrossRef]
61. Sodhi, N.; Yao, B.; Anis, H.K.; Khlopas, A.; Sultan, A.A.; Newman, J.M.; Mont, M.A. Patient satisfaction and outcomes of static progressive stretch bracing: a 10-year prospective analysis. Annals of Translational Medicine 2019, 7. [Google Scholar] [CrossRef]
62. Dunaway, S.; Dezsi, D.B.; Perkins, J.; Tran, D.; Naft, J. Case report on the use of a custom myoelectric elbow–wrist–hand orthosis for the remediation of upper extremity paresis and loss of function in chronic stroke. Military medicine 2017, 182, e1963–e1968. [Google Scholar] [CrossRef]
63. Androwis, G.J.; Engler, A.; Rana, S.; Kirshblum, S.; Yue, G.H. The Rehabilitation Effects of Myoelectric Powered Wearable Orthotics on Improving Upper Extremity Function in Persons with SCI. 2021 43rd Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC). IEEE, 2021, pp. 4944–4948.
64. Androwis, G.J.; Engler, A.; Rana, S.; Kirshblum, S.; Yue, G. Upper Extremity Functional Improvements in Persons with SCI Resulted from Daily Utilization of Myoelectric Powered Wearable Orthotics. 2021 43rd Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC). IEEE, 2021, pp. 4949–4952.
65. Hoppe-Ludwig, S.; Armitage, J.; Turner, K.L.; O’Brien, M.K.; Mummidisetty, C.K.; Koch, L.M.; Kocherginsky, M.; Jayaraman, A. Usability, functionality, and efficacy of a custom myoelectric elbow-wrist-hand orthosis to assist elbow function in individuals with stroke. Journal of Rehabilitation and Assistive Technologies Engineering 2021, 8, 20556683211035057. [Google Scholar] [CrossRef] [PubMed]
66. Pundik, S.; McCabe, J.; Skelly, M.; Salameh, A.; Naft, J.; Chen, Z.; Tatsuoka, C.; Fatone, S. Myoelectric Arm Orthosis in Motor Learning-Based Therapy for Chronic Deficits After Stroke and Traumatic Brain Injury. Frontiers in Neurology 2022, 13, 19. [Google Scholar] [CrossRef]
67. Chang, S.R.; Hofland, N.; Chen, Z.; Tatsuoka, C.; Richards, L.G.; Bruestle, M.; Kovelman, H.; Naft, J. Myoelectric Arm Orthosis Assists Functional Activities: A 3-Month Home Use Outcome Report. Archives of Rehabilitation Research and Clinical Translation 2023, 5, 100279. [Google Scholar] [CrossRef]
68. McCabe, J.P.; Henniger, D.; Perkins, J.; Skelly, M.; Tatsuoka, C.; Pundik, S. Feasibility and clinical experience of implementing a myoelectric upper limb orthosis in the rehabilitation of chronic stroke patients: a clinical case series report. PloS one 2019, 14, e0215311. [Google Scholar] [CrossRef]
69. Nilsson, M.; Ingvast, J.; Wikander, J.; von Holst, H. The Soft Extra Muscle system for improving the grasping capability in neurological rehabilitation. 2012 IEEE-EMBS conference on biomedical engineering and sciences. IEEE, 2012, pp. 412–417. [CrossRef]
70. Osuagwu, B.A.; Timms, S.; Peachment, R.; Dowie, S.; Thrussell, H.; Cross, S.; Heywood, T.; Shirley, R.; Taylor, J. Clinical Trial of the Soft Extra Muscle Glove to Assess Orthotic and Long-Term Functional Gain Following Chronic Incomplete Tetraplegia: Preliminary Functional Results. International Conference on NeuroRehabilitation. Springer, 2018, pp. 385–389. [CrossRef]
71. Kottink, A.I.R.; Nikamp, C.D.; Bos, F.P.; van der Sluis, C.K.; van den Broek, M.; Onneweer, B.; Stolwijk-Swüste, J.M.; Brink, S.M.; Voet, N.B.; Buurke, J.B.; et al. Therapeutic Effect of a Soft Robotic Glove for Activities of Daily Living In People With Impaired Hand Strength: Protocol for a Multicenter Clinical Trial (iHand). JMIR research protocols 2022, 11, e34200. [Google Scholar] [CrossRef]
72. Cousins, D.; Porto, R.; Bigelo, A.; Fox, R.; Libs, B.; Holmes, M.; Cort, J. Effects of the Ironhand® Soft Exoskeleton on Forearm Muscle Activity During in Field Automotive Assembly Tasks. Available at SSRN 4559929.
73. Hoffman, H.B.; Blakey, G.L. New design of dynamic orthoses for neurological conditions. NeuroRehabilitation 2011, 28, 55–61. [Google Scholar] [CrossRef]
74. Jeon, H.s.; Woo, Y.K.; Yi, C.h.; Kwon, O.y.; Jung, M.y.; Lee, Y.h.; Hwang, S.; Choi, B.r. Effect of intensive training with a pring-assisted hand orthosis on movement smoothness in upper extremity following stroke: A pilot clinical trial. Topics in stroke rehabilitation 2012, 19, 320–328. [Google Scholar] [CrossRef]
75. Stuck, R.A.; Marshall, L.M.; Sivakumar, R. Feasibility of SaeboFlex upper-limb training in acute stroke rehabilitation: a clinical case series. Occupational therapy international 2014, 21, 108–114. [Google Scholar] [CrossRef]
76. Andriske, L.; Verikios, D.; Hitch, D.; et al. Patient and therapist experiences of the SaeboFlex: A pilot study. Occupational therapy international 2017, 2017. [Google Scholar] [CrossRef]
77. Barry, J.; Saabye, S.; Baker, D.; Fogarty, J.; Beck, K. Effects of the saeboflex® orthosis and a home exercise program on upper extremity recovery in individuals with chronic. Journal of Neurologic Physical Therapy 2006, 30, 207. [Google Scholar] [CrossRef]
78. Lannin, N.A.; Cusick, A.; Hills, C.; Kinnear, B.; Vogel, K.; Matthews, K.; Bowring, G. Upper limb motor training using a Saebo™ orthosis is feasible for increasing task-specific practice in hospital after stroke. Australian occupational therapy journal 2016, 63, 364–372. [Google Scholar] [CrossRef]
79. Huang, Z.; Liu, J.; Li, Z.; Su, C.Y. Adaptive impedance control of robotic exoskeletons using reinforcement learning. 2016 International Conference on Advanced Robotics and Mechatronics (ICARM). IEEE, 2016, pp. 243–248. [CrossRef]
80. Ren, J.L.; Chien, Y.H.; Chia, E.Y.; Fu, L.C.; Lai, J.S. Deep learning based motion prediction for exoskeleton robot control in upper limb rehabilitation. 2019 International Conference on Robotics and Automation (ICRA). IEEE, 2019, pp. 5076–5082. [CrossRef]
81. Rose, L.; Bazzocchi, M.C.; Nejat, G. A model-free deep reinforcement learning approach for control of exoskeleton gait patterns. Robotica 2022, 40, 2189–2214. [Google Scholar] [CrossRef]
82. Côté-Allard, U.; Fall, C.L.; Drouin, A.; Campeau-Lecours, A.; Gosselin, C.; Glette, K.; Laviolette, F.; Gosselin, B. Deep learning for electromyographic hand gesture signal classification using transfer learning. IEEE transactions on neural systems and rehabilitation engineering 2019, 27, 760–771. [Google Scholar] [CrossRef]
83. Theurel, J.; Desbrosses, K. Occupational exoskeletons: overview of their benefits and limitations in preventing work-related musculoskeletal disorders. IISE Transactions on Occupational Ergonomics and Human Factors 2019, 7, 264–280. [Google Scholar] [CrossRef]
84. Elprama, S.A.; Vanderborght, B.; Jacobs, A. An industrial exoskeleton user acceptance framework based on a literature review of empirical studies. Applied Ergonomics 2022, 100, 103615. [Google Scholar] [CrossRef]