Robotic Systems for Hand Rehabilitation—Past, Present and Future

Abstract

Background: Cerebrovascular accident, commonly known as stroke, Parkinson’s disease, and multiple sclerosis represent significant neurological conditions affecting millions globally. Stroke remains the third leading cause of death worldwide and significantly impacts patients’ hand functionality, making hand rehabilitation crucial for improving quality of life. Methods: A comprehensive literature review was conducted analyzing over 300 papers, and categorizing them based on mechanical design, mobility, and actuation systems. To evaluate each device, a database with 45 distinct criteria was developed to systematically assess their characteristics. Results: The analysis revealed three main categories of devices: rigid exoskeletons, soft exoskeletons, and hybrid devices. Electric actuation represents the most common source of power. The dorsal placement of the mechanism is predominant, followed by glove-based, lateral, and palmar configurations. A correlation between mass and functionality was observed during the analysis; an increase in the number of actuated fingers or in functionality automatically increases the mass of the device. The research shows significant technological evolution with considerable variation in design complexity, with 29.4% of devices using five or more actuators while 24.8% employ one or two actuators. Conclusions: While substantial progress has been made in recent years, several challenges persist, including missing information or incomplete data from source papers and a limited number of clinical studies to evaluate device effectiveness. Significant opportunities remain to improve device functionality, usability, and therapeutic effectiveness, as well as to implement advanced power systems for portable devices.

1. Introduction

Cerebrovascular accident, commonly known as stroke, is a global health challenge that affected 17 million individuals in 2021. Stroke is the third leading cause of mortality worldwide [1] and one of the main contributors to adult physical disability [2], impacting more than a person’s health [3]. While stroke remains significant, 1.12 million cases and 9.53 million survivors of which are reported annually in the EU [4], robotic rehabilitation serves a larger patient population, including patients who suffer from Parkinson’s disease, 8.5 million cases of which are reported globally [5,6], and patients with neuromuscular conditions like multiple sclerosis, 2.8 million cases of which are reported worldwide [7] as of 2020.

After a stroke, survivors often face various complications including movement problems, speech disorders, visual disturbances, personality alterations, cognitive dysfunction, and post-stroke depression, requiring comprehensive rehabilitation approaches [8,9]. The main motor impairments are represented by spasticity and weakness [10,11], with spasticity manifesting as muscle stiffness and increased resistance to passive movements [12], while weakness affects the motor function of the fingers [13].

The effectiveness of rehabilitation relies on neuroplasticity, the brain’s capacity to change, reorganize, and adapt, and to enhance its ability to handle new situations [14,15]. Two phases are involved in neuroplasticity: the transformation of the initial neural network [16,17] and subsequent neuronal connection reestablishment [18], enhanced by physical activity and neural stimulation [19,20]. The rehabilitation process involves three main techniques [21]: passive motion involving a therapist (or the healthy hand of the patient) to guide the motion that the affected hand cannot perform anymore, active motion that involves independent movement of the affected hand to the limits of spasticity, and active-assisted therapy that combines both techniques, involving independent motion of the affected hand to the limits of spasticity and beyond this limit with the help of a therapist or patient healthy hand. Active therapy is essential for muscular strengthening and the neuroplasticity process [22,23,24,25,26].

Motor impairments can vary from individual to individual, and muscle state, stroke phase, and severity [27] must be taken into consideration. Current technology can provide precise therapeutic guidance [28], but not considering individual patient differences can reduce the effectiveness of the rehabilitation process [29]. The current growing demand for rehabilitation therapy faces significant challenges such as an insufficient availability of trained therapists [30,31] and a reduced number of therapy sessions [32,33,34].The following tools are used to measure the progress of the rehabilitation process: the CAHAI (Chedoke Arm and Hand Activity Inventory) [35,36], BBT (Box and Block Test) [37,38,39], 9-HPT (9-Hole Peg Test) [40,41], and Ad-AHA (Adult Assisting Hand Assessment stroke) [42,43], complemented by patient-reported outcomes through the MHQ (Michigan Hand Outcomes Questionnaire) [44,45], JHFT (Jebsen–Taylor Hand Function Test), which simulates daily activities for unilateral hand function assessment [46,47], DHI (Dizziness Handicap Inventory) [48,49,50,51,52], and the ABILHAND questionnaire [53,54,55].

The need for new methods to deliver care to the affected patients is reflected in the growing rehabilitation robotics market, which reached approximately USD 500 million in North America in 2020 [56], with expected growth in Asia and Europe [57]. This combination of high neurological disorder prevalence, limited healthcare professional availability, and proven benefits of consistent rehabilitation necessitates the development of hand exoskeletons to supplement traditional therapy methods and address the growing gap between patient needs and available care.

This paper presents a systematic review of rehabilitation hand exoskeletons. Section 1 presents the global impact of neurological disorders, stressing the most important ones, such as stroke, Parkinson’s disease, or multiple sclerosis, and discusses rehabilitation needs, neuroplasticity importance, and growing market demand for rehabilitation robotics. Section 2 presents human hand biomechanics, joint types, and motions, and outlines systematic review methodology including database searches, inclusion criteria, and analysis frameworks for over 300 devices. Section 3 categorizes devices into orthoses, exoskeletons, and end-effector types, discusses design requirements including safety, comfort, force transmission, and affordability, and compares mechanical approaches and clinical considerations. Section 4 classifies exoskeletons by type (soft/rigid/hybrid), mobility, mechanism position, actuation, and transmission systems. Section 5 analyzes a database of 300+ devices, presenting trends in publication, geographic distribution, design approaches, and technical specifications, and includes statistical analysis of actuator types and device configurations. Section 6 summarizes findings showing that rigid exoskeletons dominate (50%), with soft (35%) and hybrid (15.5%) following. The section also notes electric actuation prevalence (55.8%) and identifies future research opportunities for improving functionality and therapeutic effectiveness.

2. Materials and Methods

The human hand is one of the most complex and highly versatile parts of the human body, with many degrees of freedom (DoF) that allow a wide range of movement. Various types of joints contribute to the overall dexterity of the human hand; however, the large number of DoF constitutes the main challenge in creating accurate models of the hand’s biomechanics [58].

2.1. Skeletal Model of the Human Hand

The skeletal model of the human hand is illustrated in Figure 1. The OpenSim [59] library’s hand skeleton model provides insight into the hand’s structure. The hand skeleton consists of the capitate bone near the wrist, metacarpals, and phalanges. Each digit contains one metacarpal segment. The four fingers consist of three segments: proximal, intermediate, and distal phalanges. In contrast, the thumb has only two phalanx segments: proximal and distal.

The names of the hand joints are based on the bones they connect. Each of the four fingers consists of one metacarpophalangeal joint (MCP), one distal interphalangeal joint (DIP), and one proximal interphalangeal joint (PIP). The thumb consists of a MCP and one DIP joint. Near the wrist, each digit connects through a carpometacarpal joint (CMC). The joint that connects the phalanges can be considered a 1 DoF hinge joint because each pair of phalanges can only perform bending and extending movements.

The MCP joint is more complex and can be considered a 2 DoF joint. It can be modeled as a ball-and-socket joint that allows for rotation along two different directions.

The CMC joints, positioned near the wrist, can be considered 2 DoF saddle joints.

2.2. Motions of the Human Hand

Precise terminology is mandatory in the study of human hand biomechanics to describe the various movements of the digits and joints. Flexion and extension motions refer to the decrease and increase of joint angles, respectively. Abduction and adduction describe the movements of fingers away from and towards the middle finger, except for the middle finger itself, which can perform only bilateral abduction [60,61].

Due to its unique anatomical structure, the thumb possesses a more complex range of motion. The abduction of the thumb can be defined as anterior movement perpendicular to the palm, while adduction returns it to the palmar plane.

Circumduction refers to the thumb’s movement in a conical pattern at the TMC joint. Pronation represents the movement of the thumb along its longitudinal axis inward toward the index finger. The outward rotation of the thumb is termed supination. The terminology described above provides an accurate description of finger movement. Finger motions, flexion, extension, abduction, adduction, and circumduction are illustrated in Figure 2.

2.3. Predominant Human Hand Postures

The main movements of the human hand are illustrated in

Table 1. Predominant human hand movements.

Movement	Illustration of the Movement	Movement	Illustration of the Movement
Thumb up		Extension of index and middle, flexion of the other fingers
Flexion of ring and little finger extension of the other fingers		Flexion of fingers except thumb
Abduction of all fingers		Fingers flexed together in fist
Pointing index		Thumb opposing base of little finger

and are in-depth analyzed in reference [62].

The most common grasp movements of the human hand are depicted in

Table 2. Common human hand grasping gestures.

Grasping Movement	Illustration of the Grasping Movement	Grasping Movement	Illustration of the Grasping Movement
Large diameter grasp		Tripod Grasp
Small diameter grasp		Prismatic pinch grasp
Fixed hook grasp		Tip pinch grasp
Index finger extension grasp		Quadpod grasp
Medium wrap		Lateral grasp
Ring grasp		Parallel extension grasp
Prismatic four fingers grasp		Extension type grasp
Stick grasp		Power disk grasp
Writing tripod grasp		Open a bottle with a tripod grasp

and in-depth analyzed in reference [62].

2.4. Data Extraction and Categorization

The protocol illustrated in Figure 3 was used to systematically identify and analyze relevant research articles focusing on hand exoskeleton devices for rehabilitation and assistance.

From each included article, a wide range of data were extracted. This included bibliometric information such as authors, year of publication, country of origin, and publication type (journal article or conference paper). Technical details extracted included the device name (if applicable), type of exoskeleton (rigid, soft, or hybrid), actuation method and type, transmission mechanism, number of actuators, degrees of freedom (total and active), mechanism placement, finger motions assisted, independent vs. coupled finger assistance, number of fingers covered, and range of motion. Additional information on target application, portability, safety features, adaptability features, weight, and any clinical testing was also recorded.

2.5. Literature Search and Selection

To identify relevant studies on hand robotic exoskeletons for rehabilitation and assistance, a comprehensive literature search was conducted. The search covered multiple databases including PubMed, IEEE Xplore, Scopus, Web of Science, and Google Scholar, using search terms such as “hand exoskeleton” and “hand rehabilitation robot”.

The extracted data was organized into a structured database for analysis as illustrated in Figure 4. Devices were categorized based on their design characteristics, actuation methods, and intended applications, facilitating comprehensive analysis.

Initially, more articles were identified. After removing duplicates and screening titles and abstracts, articles were selected for full-text review. Following the application of inclusion and exclusion criteria, a final set of 306 articles was included in this review. Additionally, over 20 existing review articles in the field were examined to identify any potentially missing relevant studies.

The inclusion criteria for the review encompassed studies describing hand exoskeleton devices for rehabilitation or assistance, conference proceedings, and articles. Exclusion criteria included studies focused only on upper limb exoskeletons without specific hand components.

2.6. Data Analysis

The analysis of the extracted data involved quantitative approaches. The quantitative analysis utilized descriptive statistics to summarize trends in hand exoskeleton design and functionality. This included examining the distribution of exoskeleton types, distribution of different actuation methods, distribution of transmission mechanisms, average number of DoF, distribution of finger coverage, assisted motions, and weight distribution analysis. Trends in publication types and research output over time, as well as the geographical distribution of research, were also analyzed.

2.7. Limitations

Several limitations of this review were identified. The focus on English-language publications may have excluded relevant studies published in other languages. Despite efforts to be comprehensive, it is possible that some relevant studies were missed due to the specific search terms used or limitations of the databases searched.

3. Hand Rehabilitation Devices

Hand rehabilitation devices (HRDs) can be classified into three main categories: orthoses, exoskeletons, and end-effector devices. Orthoses, similar to traditional hand braces, provide fundamental support. Exoskeletons and end-effector devices go a step further by incorporating powered systems that enable passive exercises. These advanced devices offer crucial support and movement of the fingers through various actuation methods, adding an important dimension to rehabilitation therapy [21,22]. Robot-assisted therapy for upper extremity rehabilitation post-stroke emerged in the 1990s [63] the initial hand rehabilitation robots were designed to take over the physically demanding aspects of therapy from human therapists. The main objectives were to better reproduce the natural motion patterns of human joints and more effectively address the specific needs of rehabilitation. This approach developed better and more flexible robotic systems in hand rehabilitation, leading to the modern devices used today [64].

The field of rehabilitation robotics integrates multidisciplinary knowledge fields, such as anatomy, neuroscience, cognitive and learning sciences, and extensive clinical experience [65].

The development of hand rehabilitation robots requires more than mechanical and engineering considerations; the characteristics of stroke patients, current rehabilitation theories, and methods for assessing therapeutic progress should be taken into consideration [66,67,68,69,70,71,72,73,74].

Based on mechanical structure, hand rehabilitation devices can be classified into two main categories, exoskeleton devices and end-effector devices.

• Exoskeleton Devices

Exoskeletons are devices that wrap around the target body part, similar to orthoses in their ability to provide safety and support. However, exoskeletons are distinguished by their additional powered components. These extra elements enable exoskeletons to offer active, powered functions to the user. This combination of supportive structure and powered assistance makes exoskeletons particularly useful for rehabilitation, movement augmentation, or other specialized applications where active support is beneficial [21,75].

Finger exoskeletons enable precise actuation of specific joints while providing support, which is crucial for patients with advanced injuries. Their portability makes them valuable for elderly or remote patients, offering consistent rehabilitation or assistance outside clinical settings [75].

• End-Effector Devices

End-effector devices offer a distinct approach to hand rehabilitation compared to exoskeletons. These devices focus on controlling the endpoint or the distal interphalangeal joint of the finger, rather than encompassing and actuating the entire finger structure.

The primary advantages of end-effector devices are their high level of control and feedback precision. By concentrating on the fingertip or the most distal joint, these devices can achieve accurate movement and force application without the complexity of managing multiple joints simultaneously.

4. Exoskeleton Rehabilitation Devices

Exoskeletons are designed to match the shape and the movement of the user’s hand. Each degree of freedom in the exoskeletons’ design must be precisely aligned with the corresponding human joint. By mimicking human anatomy and joint movements so closely, exoskeletons can provide targeted support and assistance, making them a valuable tool in robotic rehabilitation [75].

Exoskeleton rehabilitation devices can be classified as follows.

4.1. Classification by Type

Exoskeleton hand rehabilitation devices can be divided into three main categories: rigid, soft, and hybrid, as illustrated in Figure 5.

4.1.1. Soft Exoskeletons

In recent years, the development of soft robotics rehabilitation systems based on soft actuation has increased significantly [76].

Soft exoskeletons, sometimes referred to as soft gloves [77], offer unique advantages in rehabilitation. These soft exoskeletons can reduce the overall weight of hand rehabilitation systems and provide better self-alignment between the device and finger joints due to their soft actuation [78,79]. These exoskeletons are typically manufactured using fabric-, plastic-, or silicone-based materials. This flexibility allows the exoskeleton to be customized for each patient’s finger size, offering a more personalized rehabilitation experience [78].

4.1.2. Rigid Exoskeletons

Rigid exoskeletons utilize rigid transmission mechanisms, which are systems of rigid links joined together [80]. This is one of the simple ways to translate the movements of the actuator to the intended human joint.

In systems that are fully actuated, both direct and indirect relationships can be established. Using this feature, any measurements taken from the actuator, such as encoder signals, torque, or current, can be used to determine the finger’s state. Simple and natural control of the hand is possible due to the direct connection between actuators and hand joints and, as a result, additional tracking systems are not required.

Control algorithms are simplified due to the natural matching of actuator and joint movements, and the overall reliability and precision of the exoskeleton are enhanced. Rigid exoskeletons enable more accurate position control and force transmission, which are crucial for both rehabilitation exercises and assistive applications [76].

Based on the linkage type, the rigid hand exoskeletons can be classified as illustrated in Figure 6. Figure 7 illustrates a schematic representation of each linkage type.

Hand exoskeletons can be developed with mechanisms that create remote centers of motion for each link, aligning with the finger joints’ natural pivot points, as illustrated in Figure 7a. This type of mechanism has been explored with various potential designs discussed in reference [81]. Popular mechanisms include the parallelogram mechanism and circular prismatic joints [82].

Matched axis types of exoskeletons aim to align joints precisely with those of the human hand, as illustrated in Figure 7b. This method is one of the simplest ways to control hand movements predictably; however, the design presents challenges to implement in practice. The main reason is represented by the placement of the exoskeleton’s joints directly next to the finger joints. This type of placement is problematic for multiple fingers due to interference with natural hand movements when fingers need to come close together during grasping. Due to these practical limitations, the matched axes topology is generally not considered a viable option for exoskeletons designed to support multiple fingers simultaneously [82].

An alternative method for connecting the movement of a specific exoskeleton joint to a single hand joint is represented by a redundant linkage. This topology involves the attachment of the system to the digit at points before and after each actuated joint. However, additional linkages and joints should be incorporated between attachment points, as illustrated in Figure 7c. Well-defined relationships between the motions of the exoskeleton and the hand joints are established using this type of mechanism [82].

Multiple finger joints can be controlled using a single actuator through an underactuated mechanism. The main feature of this design is represented by the ability to automatically adjust the forces that act on the different finger phalanges based on the interaction between the device and the user’s finger. The adaptive behavior of the device is possible due to the incorporation of passive elements in the mechanism design [72], as illustrated in Figure 7d.

Coupled exoskeleton devices coordinate the movement of different finger joints simultaneously and are designed to interact with the user’s finger at multiple contact points, as illustrated in Figure 7e. This type of design allows for more natural finger movement. Through mechanical adjustments of the ratios, these devices can accommodate different motion patterns and adapt to various users without complex control systems [72].

Another type of exoskeleton linkage is represented by the fingertip exoskeleton. Using this type of mechanism, the fingertip’s position in space is controlled, regardless of how the individual finger joints move. As illustrated in Figure 7f, this design allows precise fingertip manipulation while enabling natural movement of the finger joints [72]. This type of mechanism is simpler than multi-joint mechanisms. Figure 8 provides visual representations of various linkage types.

Prior investigations into redundant linkage exoskeletons include the works of [86,87,88,89,90,91]. Additional research exploring fingertip linkage devices has been conducted by several authors [92,93,94,95,96]. Investigations into underactuated devices have been studied in [97,98]. Understanding human hand proportions and exoskeleton connectivity has been explored in [99,100].

A considerable number of studies have investigated coupled devices utilizing a single actuator, for example, the work of [101,102,103,104,105,106]. Another strand of research has focused on coupled devices employing two actuators [107,108,109,110,111]. The research conducted by [81] introduced an exoskeleton design utilizing a remote center of motion (RCM) mechanism.

4.1.3. Hybrid Exoskeletons

Efficient power transmission, robust force control, and precise kinematics are advantageous characteristics provided by rigid exoskeletons. However, excessive mass, bulky design, and high stiffness are some of the main disadvantages of this type of rigid exoskeleton. To address the disadvantages, researchers developed innovative approaches such as the incorporation of elastic elements within the rigid mechanism to enhance compliance while preserving the advantages of structural rigidity. Hybrid exoskeletons integrate a rigid frame with deformable materials to achieve an optimal balance between mechanical performance and ergonomic consideration [76].

The integration of compliant materials in exoskeleton design significantly enhances device wearability and user comfort. These materials facilitate compatibility with the natural biomechanics of human fingers, improving force output capabilities [112]. Additive manufacturing technologies have facilitated the development of diverse exoskeleton architectures [113]. Some researchers have leveraged these capabilities to create single-unit designs [114,115,116] while others have explored modular structures [117,118]. An alternative strategy involves the incorporation of elastic spring elements [119]. This approach aims to mitigate the limitations typically associated with rigid structures while preserving their mechanical advantages. Another strategy employs fiber-reinforced soft bending actuators [120,121,122].

4.2. Classification by Hand Mobility

Exoskeletons can be classified according to hand mobility. Devices can be designed to assist and control different numbers of fingers on the human hand, which typically has five fingers. Figure 9 illustrates various configurations of finger assistance in exoskeleton designs.

Hand exoskeletons can be categorized by the number of actuated fingers. Single-finger designs [13,83,85,95,97,102,103,118,123,124,125,126,127,128,129,130,131,132,133,134,135,136,137,138,139,140,141,142,143,144,145,146,147,148,149,150,151,152,153,154,155], often focusing on the index finger [13,83,102,118,123,124,144,151,152], serve as initial investigations for multi-finger exoskeletons. Other single-finger exoskeletons target the thumb [142,150,156] or middle finger [133].

Two-finger exoskeletons [157,158,159,160,161,162,163,164,165,166,167,168,169,170], which independently control multiple finger configurations such as thumb–index [103,158,164,167] or index–middle [159,171], have been developed to support targeted hand movements for rehabilitation, such as precision grip tasks. However, most activities of daily living require assistance from at least three fingers [96,98,110,172,173,174,175,176,177,178,179,180,181,182,183,184]. Three-finger combinations typically focus on thumb–index–middle configurations [175,176,182,183,185,186].

While three-finger designs can support many tasks, configurations of four fingers [84,88,89,91,112,187,188,189,190,191,192,193,194,195,196,197,198,199,200,201,202,203,204] offer improved functionality. These include designs assisting all fingers except the thumb [84,106,205], configurations excluding the little finger [189,201,206], and systems actuating thumb–index–middle–ring combinations [89].

Most of the full hand (five fingers) exoskeletons employ independent actuation of all digits [15,93,117,119,120,146,207,208,209,210,211,212,213,214,215,216,217,218,219,220,221,222,223,224,225,226,227,228,229,230,231]. These systems enable individual control of each finger, offering maximum dexterity for both rehabilitation and assistive functions. While such designs typically require more complex control systems and a higher number of actuators, they provide the flexibility needed for precise manipulation tasks and personalized rehabilitation strategies.

Some devices implement coupled actuation designs [15,84,109,114,117,119,196,232,233,234,235,236,237,238,239,240,241,242,243,244,245,246], where certain fingers are mechanically linked to move together. Common coupling strategies include grouping four fingers together with independent control of the thumb [15,114,210,233,235,245,247], using ring–little finger coupling while maintaining independent control of the other digits [232,248,249], or coupling all fingers into a single actuation unit [237,238,250,251]. Some designs implement different coupling combinations for flexion and extension movements [242,243,244,252], offering a balance between functionality and complexity. These approaches offer advantages in terms of mechanical simplicity and a reduced number of actuators.

4.3. Classification by Mechanism Position

Robotic exoskeletons can be categorized based on the placement of the mechanism. This classification includes palmar, lateral, dorsal, and glove-type configurations, as illustrated in Figure 10. Each placement modality offers distinct biomechanical advantages and constraints, influencing the exoskeleton’s functionality [72].

A schematic representation of each exoskeleton configuration is depicted in Figure 11, illustrating the palmar, lateral, dorsal, and glove-type designs.

4.3.1. Palmar

Palmar devices are characterized by the placement of mechanical or transmission components within the palmar region of the hand, as illustrated in Figure 11a. This type of mechanism placement has been studied in papers [190,253,254,255,256]. However, this configuration presents a limitation for assistive applications.

4.3.2. Lateral

Lateral devices are characterized by the positioning of mechanical or transmission components along the lateral aspects of finger phalanges, as depicted in Figure 11b, and were studied in [85,90,103,129,132,157,158,201,209,234,237,239,256,257,258]. An important advantage of lateral devices is the preservation of palmar surface availability for tactile interactions with the environment. However, lateral configurations present certain limitations; in multi-finger implementations, there is a potential for inter-digit collisions [72].

4.3.3. Dorsal

Dorsal devices are characterized by the placement of mechanical transmission above the finger phalanges, as illustrated in Figure 11c [59,83,84,85,87,88,89,90,91,92,93,94,95,96,98,101,102,105,106,107,108,109,110,111,112,113,114,117,120,123,124,125,126,127,128,130,131,134,135,136,137,139,140,141,142,143,144,145,146,147,148,149,150,151,153,154,156,158,160,161,162,163,164,166,169,173,181,182,183,185,188,195,196,197,198,199,200,201,202,203,204,205,206,207,210,211,212,213,214,217,219,220,221,222,224,225,226,227,230,233,235,236,238,240,242,244,245,247,250,251,259,260,261,262,263,264,265,266,267,268,269,270,271,272,273,274,275,276,277,278,279,280,281,282,283,284,285,286,287,288,289,290,291,292,293,294,295,296,297,298,299,300,301,302,303,304,305,306,307,308,309,310,311,312,313,314,315,316,317,318,319,320,321,322,323,324,325,326,327,328,329,330,331,332,333,334,335,336,337,338,339,340,341,342,343,344,345,346,347,348,349,350,351,352,353,354,355,356,357]. This configuration minimizes inter-digit collisions while preserving palmar surface availability.

4.3.4. Glove

Glove exoskeletons depicted in Figure 11d are characterized by their flexible, fabric-like construction that envelops the hand, with actuation components integrated into or attached to the glove material. This type of design offers several advantages. The soft, pliable nature of these devices allows for extended wear without causing discomfort or restricting natural hand movements. Soft gloves can conform to various hand sizes and shapes, making them suitable for a wider range of users without extensive customization. By using thin, flexible materials, soft gloves preserve much of the user’s natural tactile sensation, allowing for better interaction with objects during rehabilitation exercises.

Soft glove exoskeletons have been the subject of extensive research and development in recent years, with significant contributions from various studies [121,133,173,174,175,176,180,184,186,189,191,192,193,222,228,229,240,243,248,249,262,358,359,360,361,362,363,364,365,366,367,368,369,370,371,372,373,374,375,376,377,378,379,380,381,382,383,384,385,386,387,388]. These investigations have explored diverse aspects of glove design, fabrication, and implementation, advancing the field of wearable rehabilitation technology. Figure 12 illustrates each type of mechanism placement with examples from the literature.

4.4. Classification by Actuator Type

Exoskeleton actuation modalities can be classified into electromechanical, pneumatic, shape memory alloy-based, hydraulic, contralateral extremity-driven, human muscle, series elastic actuators, and other approaches as depicted in Figure 13.

4.4.1. Electric Motors

Exoskeletons that use rotary DC motors as their actuation method have been the subject of extensive research in recent years, and numerous studies have explored various aspects of implementing this type of motor in exoskeleton systems [13,15,83,87,90,92,93,94,95,96,97,101,103,106,108,110,121,122,123,125,126,128,129,130,133,135,136,137,138,139,140,141,142,143,154,157,158,160,161,162,164,167,170,173,175,184,185,186,189,192,194,195,202,204,208,211,227,228,231,234,237,239,240,244,247,248,250,251,255,256,261,262,263,264,273,279,306,308,311,321,324,338,340,361,365,366,370,376,384,387,389,390,391,392,393,394,395]. Linear motors are a good choice for controlling finger movements in exoskeleton hands. They can be easily placed on top of the hand and can open and close multiple fingers together [15,59,84,98,104,112,117,132,144,145,146,150,156,159,171,180,196,197,200,207,212,213,214,217,231,232,233,253,260,269,281,283,284,285,286,288,299,310,331,367,368,372,376,383,396,397]. Servomotors represent another type of electric actuator that is used in exoskeleton design. These types of motors offer precise position control and rapid response, making them suitable for accurate joint angle manipulation. Several researchers used servomotors in their exoskeleton prototypes, as documented in various publications [84,91,107,119,127,144,145,171,174,188,190,197,205,212,229,235,254,280,282,283,284,285,286,287,309,325,337,339,396,398].

4.4.2. Pneumatic Actuators

Pneumatic actuators [132,134,152,153,163,180,191,193,198,201,203,206,217,219,221,222,224,225,241,243,249,252,257,270,271,272,289,290,291,292,293,294,312,313,316,318,319,323,326,328,330,342,358,359,360,363,364,371,372,373,374,375,377,378,380,382,385,386,388,399,400,401,402] are widely used in hand rehabilitation robots due to their advantages in force generation, control, and weight-to-torque ratio [214,288]. They operate using air flow through various mechanisms such as pneumatic cylinders [102,109,187], air balloons [238], air bladders [241,359], flexible thermoplastic fabrics [249,360], or pneumatic artificial muscles [88,124,203,262,272,318,330,402]. Different categories of soft pneumatic actuators have been developed. Pneu-nets soft actuators (PNSAs) [198,221], made of elastomers, move finger joints directly. Optimized designs use semicircular chambers for better performance [217]. Fiber-reinforced actuators generate higher forces compared to PNSAs and allow for better designs [163,220,289,290,403]. Positive-negative pneumatic actuators allow active control of both flexion and extension [291,292]. Soft tissue-based pneumatic actuators use flexible, lightweight tissues and can achieve high forces [252,293]. McKibben-type pneumatic artificial muscles, made of a rubber inner tube with a braided shell, inflate and contract when pressurized [404]. Manufacturing processes for these actuators include casting in molds, direct 3D printing, and monolithic molding techniques [134].

The main advantages of pneumatic actuators include high, adjustable force and speed at a low cost, good weight-to-torque ratio [214], less maintenance [66,405,406,407], and the ability to stop under load without damage [69,405]. Some alternative approaches include EMG-controlled systems [226] and various mechanical design configurations [408], each offering distinct benefits for rehabilitation applications.

However, pneumatic actuators have several limitations. These include size constraints due to compressors and air storage chambers, noise, control difficulties due to time variability and nonlinearity, and slower response times in some cases [29,32].

Other studies on pneumatic actuation explored various designs for hand exoskeletons, for example, the work of [295,380]. These designs include soft actuators [198,218], fabric-based systems [216,249,252,293,312,400], and fiber-reinforced silicone actuators [380]. Specialized approaches like thin McKibben muscles [401] and bending-type pneumatic muscles [294] were also investigated. Research has focused on soft exoskeletons for rehabilitation and assistance [193,271,272,358,364,371,374,378] and incorporating various materials such as silicone rubber [134,152,153,163,191,193,206,217,219,220,224,225,290,291,292,313,316,319,328,342,358,360,363,364,371,373,375,377,380,385,390], flexible 3D-printed materials [270,333,374], and custom pneumatic actuators [270,271].

4.4.3. Shape Memory Alloys

SMAs have two key properties: shape memory effect and super elasticity [409,410]. The shape memory effect is the property of recovering the original shape upon heating to a critical temperature when it is deformed in the low-temperature phase [411]. SMAs work on the principle of phase transformation between two crystalline structures: martensite, the low-temperature phase, and austenite, the high-temperature phase [412]. After plastic deformation, SMA wire can return to its original shape when heated to a certain temperature. Additionally, under isothermal conditions, it can undergo great deformation, consume and absorb mechanical energy, and return to its original shape after removing the load, demonstrating a good damping effect [409,410].

Common materials used for SMAs include Ni–Ti and Cu–Al–Ni, among other combinations. The actuation is typically achieved by heating the SMA wire or rod by applying an electric current through it.

SMAs have been explored by several researchers [296,297,298,381,413,414,415] and have several advantages, including a high power-to-weight ratio [266]. This makes them suitable for a wide range of applications as both actuators and sensors.

SMA actuators also present several challenges. Their output motion is characterized by hysteresis, high nonlinearity, and saturation, making precise control difficult [416]. Additionally, SMA wires or rods produce significant heat during operation, thus safety considerations must be implemented when used in exoskeleton systems.

4.4.4. Hydraulic Actuators

Hydraulic actuators operate on a principle similar to pneumatic systems, but they use an incompressible liquid instead of air [277]. These actuators are known for their excellent performance characteristics [32]. Various hydraulic devices have been explored for power transfer in exoskeleton applications, including hydraulic cylinders [417], flexible inflatable materials [265], and certain types of artificial muscles [362]. However, hydraulic actuators have some limitations that affect their use in certain applications. The primary drawback is the requirement for a wider space to accommodate the liquid-transmitting pipes [32]. This spatial requirement makes hydraulic actuators less commonly used in hand rehabilitation robots [418].

4.4.5. Human MUSCLE

An innovative approach involves harnessing the power of the patient’s own muscles. This method, which can be broadly categorized as a form of actuation, utilizes functional electrical stimulation, FES, to activate impaired hand muscles [417]. This hybrid approach combines the benefits of electrical muscle stimulation with robotic assistance, potentially improving the effectiveness of hand rehabilitation. A notable example of this combined strategy is the FES and robotic glove system developed by Rong et al. [419].

4.4.6. Contralateral Extremity

In the field of hand rehabilitation, another approach involves utilizing the unaffected limb as a form of actuation for the impaired hand. This method, known as contralateral extremity actuation, is particularly relevant in rehabilitation systems that employ bilateral training strategies [419]. There are two primary ways in which the healthy extremity can drive the rehabilitation of the impaired hand. The first, which remains largely theoretical, involves direct force transfer from the unaffected hand to actuate devices on the impaired side. The second, and more researched approach, uses signals or data from the healthy hand to indirectly control rehabilitation devices on the affected side [420].

4.4.7. Series Elastic Actuators

Series elastic actuators (SEAs) represent an innovative approach to exoskeleton design, combining the power of electric motors with the compliance of spring elements. Unlike traditional actuators that prioritize joint stiffness, SEAs intentionally reduce it, opening new possibilities in rehabilitation robotics [165]. The incorporation of a spring element between the motor and the load in SEAs offers several advantages such as enhanced shock absorption and improved force control accuracy and stability. While SEAs seem to present an ideal solution for hand exoskeletons, they come with some disadvantages; the intentional reduction in joint stiffness limits the magnitude of the force that can be transferred to the exoskeleton and mechanical complexity. Some researchers have explored the potential of SEAs in hand rehabilitation devices [165,199].

4.4.8. Other Types of Actuators

Research into alternative actuator technologies continues, as engineers and researchers develop more efficient and effective solutions. Among the emerging solutions being explored are electroactive polymers and ultrasonic motors [421].

Figure 14 illustrates different types of hand exoskeletons using various actuation methods.

4.5. Classification by Transmission Type

An alternative classification method is based on power transmission mechanisms. This approach categorizes systems into several types: geared systems, linkage mechanisms, cable-driven systems, compliant structures, and restorative springs. Figure 15 illustrates these various transmission types.

4.5.1. Linkage

Linkage-based devices are a popular choice in hand rehabilitation robotics. These devices use mechanical links to form finger components and can be categorized based on their control mechanisms and structural designs.

Linkage-based actuation offers several advantages. It can be implemented with various control methods. By combining actuators with rigid exoskeletons, these devices can secure the range of motion and prevent out-of-plane flexion or extension [422,423].

This type of device also presents challenges. Motors and translational linkages can add weight to the system, even for simple motions, and compact designs require more expensive actuators [424,425,426]. The addition of linkages complicates the exoskeleton’s kinematics, with each link’s mass, inertia, and center of gravity changing for different hand sizes. Simplified designs with fewer actuators limit mobility and task adjustability.

Several studies have expanded our understanding of linkage-based devices [59,83,84,85,87,90,91,94,95,96,97,98,110,137,139,140,144,146,147,149,150,151,156,159,160,162,166,170,171,200,207,211,212,214,233,236,245,247,250,260,269,273,280,281,282,286,287,288,297,299,300,301,305,306,309,310,321,322,332,376,392,396,399].

4.5.2. Gears

Gear-motor actuation, while often considered a direct approach, can be expensive and challenging to implement. With a human hand, several gears and gear trains would be used to provide actuation. Designs that use gear-motor actuation face similar weight issues experienced by linkage-based actuation.

There is a fine line between direct drive and gear assembly. Some devices [279,311] are classified as geared systems because the motor is connected through a gear train to the structure. The idea behind a direct drive or geared system is for the actuators to be as close to the structure as possible, typically mounted on the dorsal side of the hand. This type of transmission enables the devices to be portable and less complex, simultaneously minimizing the power or transmission losses, but increasing the weight on the hand.

4.5.3. Cable

Cable-driven exoskeletons [13,15,88,93,120,124,127,130,131,133,138,154,158,161,164,167,168,173,174,175,176,178,179,180,181,182,184,185,186,189,190,195,197,204,205,208,209,213,222,223,230,234,235,237,240,242,244,248,251,254,255,256,277,284,285,296,298,317,325,327,329,334,335,337,338,339,340,365,367,368,369,370,372,381,383,384,386,387,390,393,394,395,397,427,428,429] are a popular choice for hand rehabilitation, using servo or rotary motors. These systems utilize cables to achieve low-weight solutions. The main advantage of cable-driven systems is their ability to shift the weight of actuation components away from the hand, resulting in lighter devices. Cable mechanisms can be categorized into two types: Bowden cable-driven devices and tendon-driven devices [256]. Bowden cable transmission involves routing cables or rigid rods through low-friction tubes to connect the actuating unit with the hand exoskeleton. This allows for remote placement of the actuating unit, making it suitable for stationary rehabilitation or wheelchair-mounted systems [160,251,256,275,393,430,431]. Another advantage of cable-driven systems is their use of soft exoskeletons, which provide an adaptable design that can adjust to various hand shapes and sizes [432]. This flexibility is particularly beneficial given the diverse dimensions of patients’ hands.

Cable-driven systems present several challenges; one major disadvantage is the loss of power and control issues; the actuation experiences transmission losses as the device performs exercises, mainly due to friction.

The main difference between tendon and Bowden cable systems is that tendon cables mimic the hand anatomy structure by flexing and extending the fingers with a cable routed in a glove, whereas Bowden cables use a rigid structure to transfer the cable forces to the fingers, as mentioned previously. Tendon cables are usually connected to the distal part of the finger and flex or extend the finger by applying tension to the cable. This type of transmission is normally unidirectional, meaning only one cable can execute a single task such as flexion or extension. A second cable is required to have a bidirectional actuation, which means another actuator is generally required. These types of developed hand exoskeletons are usually underactuated systems where a single actuator can flex or extend multiple fingers such as in [56,57,58] using a single cable routed in a loop across or between the fingers.

Tendon cables can suffer from cable breakage, and the routing paths are sometimes complex and can cause power losses due to friction, but they can apply the required forces to the digit itself instead of the joints.

Tendon actuation has been explored and implemented in other numerous studies, as evidenced by the research presented in [34,35,133,173,175,176,177,179,182,183,186,189,190,208,222,234,240,242,248,296,298,325,327,365,366,367,370,383,386,390,397,428,433,434,435]. Despite these challenges, cables remain an attractive option for hand rehabilitation robots due to their similarity to hand muscles [158,376].

Bowden cable mechanisms have been studied in the following studies [154,159,164,181,202,223,227,254,308,329,335,337,394,395]. Bowden cables offer more flexibility due to their cable conduit but suffer from variable and high friction forces caused by curves [125].

Pulley cables require continuous tension to maintain traction, limiting their use [128,436]. Examples of devices using pulley cables include the work of [88,235,251,393].

4.5.4. Restorative Springs

Springs, despite being passive components, are commonly integrated into hand exoskeletons that primarily use unidirectional actuation. They are often paired with either tendons or pneumatic cylinders, which typically control only flexion or extension. The springs serve to passively facilitate the opposite motion, complementing the active actuator. Numerous studies have explored and implemented spring-based actuation mechanisms in hand exoskeletons [83,197,232,274,276,284].

4.5.5. Compliant Transmission

Artificial muscles and flexible jointless structures represent innovative approaches in the field of soft robotics, offering compliant and adaptable solutions for finger flexion and extension [437].

Artificial muscles, typically driven by pneumatic or hydraulic systems, consist of a flexible material encased in a braided sheath. When fluid is injected, the muscle contracts, and when pressure is released, it extends [201,203,257,265,272,318,319,330,402]. These systems provide a high degree of flexibility [438]. However, they can face challenges such as ballooning, where excessive pressure causes the flexible material to expand beyond its functional range. Researchers have addressed this issue through careful selection of braiding materials and design optimizations [201,265].

Flexible jointless structures represent an evolution of the artificial muscle concept. These devices, often made from silicone rubber, are molded in specific configurations to achieve bending motions without the need for a braided sheath [302,360]. These structures offer the advantages of simplicity and potentially lower manufacturing costs, as they can be custom-designed for individual users using relatively inexpensive materials.

Further research on flexible jointless structures has been extensively conducted, as evidenced by numerous studies, for example [72,134,153,163,191,193,198,206,217,219,220,221,225,270,290,292,313,316,328,342,358,360,362,363,364,371,373,374,375,377,380,385]. These investigations have significantly contributed to the advancement of this technology, exploring various design configurations, materials, and applications in the field of soft robotics for rehabilitation.

Figure 16 illustrates different types of hand exoskeletons using various transmission systems.

Hand exoskeletons use a wide variety of transmission combinations to achieve optimal performance and functionality. Researchers often combine multiple transmission elements to leverage the strengths of each and address the complex requirements of hand rehabilitation and assistance.

Hand exoskeletons utilize various transmission combinations like cable and steel wire systems [119]. Some designs incorporate cable and spring mechanisms [127] while others explore linkage and spring combinations [281].

The linkage and cable combination appears to be particularly popular among researchers [91,102,103,106,109,118,123,125,130,135,136,141,142,143,148,157,159,160,169,194,196,202,227,231,259,261,278,303,391].

5. Results and Discussion

A comprehensive database was developed to catalog and analyze hand rehabilitation devices reported in the literature. The database contains over 300 devices, each of them characterized by 45 criteria. The criteria were defined in such a way to cover the technical specifications of each device.

5.1. Database Parameters Organization

To facilitate the analysis of the papers, the database was structured into six primary categories. The bibliometric information category contains fundamental publication data including author details, publication year, and article identification.

The database core structure is represented by technical specifications that include information about actuation systems, transmission mechanisms, motion characteristics such as DoF, and range of motion.

Another significant category is represented by control and sensing criteria that include the implementation of different sensors, control methods, intention detection methodologies, feedback, and the interface between the device and the patient.

Each device was documented with its practical features, covering portability, safety, weight, and how well it could adapt to different uses.

Regarding the validation of each device, implementation details were recorded such as test status and commercial availability. The information can be used to evaluate the transition from research prototype to practical application.

5.2. Database Population Process

Each device was analyzed across all six primary categories, with careful attention to maintaining uniform terminology. Missing information was explicitly noted to maintain database integrity and facilitate future updates.

The information organization enables easy access to specific device characteristics, while the structure supports detailed examination of individual devices.

5.3. Limitations and Constraints

During the development of the database, several limitations were identified such as incomplete reporting in original sources, variations in measurements, and incomplete information regarding all criteria that were used.

5.4. Discussions

The systematic data visualization conducted in this study covers design characteristics, technological implementations, and performance metrics across different hand exoskeleton configurations. The following graphs explore these findings in detail.

Figure 17a illustrates the temporal distribution of publications related to hand exoskeleton rehabilitation devices from 2002 to 2024, grouped in 5-year intervals. An important peak research output was observed in the 2016–2020 interval. This trend suggests a growing interest and investment in the field.

Figure 17b provides a concise overview of overall distribution of publication types across all years. Journal articles account for 56% of total publications while conference papers account for 44% of total publications.

Figure 18 illustrates the top ten countries that contribute to hand exoskeleton device research based on their publication output. The United States leads, followed by China and Italy.

Figure 19 illustrates the evolution of publication types across time. In the early years of the study period, conference papers were more prevalent, but a shift towards journals can be observed in recent years.

The distribution of hand exoskeleton applications across categories is illustrated in Figure 20. Rehabilitation made up 66% of the total, followed by devices used for both rehab and assistance.

The distribution of primary actuation types used in hand exoskeletons is illustrated in Figure 21. The chart categorizes actuation mechanisms into four main groups: electric, pneumatic, other, and not specified.

Electric actuation represents 55.8% of the studied exoskeletons, followed by pneumatic actuation and other types of actuation. A notable “other” category stands for 12.9% of the total studied exoskeletons.

Electric actuation represents 55.8% of the studied exoskeletons, followed by pneumatic actuation and other types of actuation.

Figure 22 breaks down the types of electric motors found in hand exoskeletons. DC motors are the predominant choice, followed by linear actuators and servomotors. Unspecified electric motors account for 8.3% of the actuators used.

Figure 23a illustrates the distribution of the main transmissions in hand exoskeleton designs. Combined mechanisms stand in the first place, followed by cable-based solutions, compliant, and linkage-based solutions. Figure 23b describes the combined transmission systems. Cable–linkage combinations are the most frequently used solution, followed by cable–spring, linkage–spring, and gear–linkage configurations.

Figure 24 breaks down the different types of hand exoskeleton designs. The study categorized exoskeletons into three main categories: rigid, soft, and hybrid. The distribution reveals that rigid designs dominate the field, accounting for approximately 50% of analyzed exoskeletons; soft designs represent a significant proportion of the dataset, comprising 35.0% of the total studied hand exoskeletons. This presence reflects the growing interest in soft robotics. Hybrid designs combine elements of both rigid and soft structures and constitute 15.5% of the studied exoskeletons.

Figure 25 illustrates the distribution of the number of actuators used. The chart reveals a various range of actuator configurations; 29.4% utilize five or more actuators, while 24.8% use only one to two actuators, indicating the prevalence of underactuated, coupled mechanisms or hand exoskeletons that are used for a limited number of fingers. A substantial percentage of 33.9% of the studied papers did not specify the number of actuators that were used.

In Figure 26, the distribution of mechanism placements is illustrated. Dorsal placement constitutes the majority of the studied hand exoskeletons, followed by glove-based, lateral, and palmar configurations.

Figure 27a presents the types of finger motions assisted by the studied hand exoskeleton designs. Most designs focus on both flexion and extension, aiming to provide a full range of motion. Figure 27b illustrates the distribution of independent vs. coupled finger assistance in studied hand exoskeleton designs. A significant majority implement independent finger actuation, allowing for more precise and versatile hand movements. Coupled designs and partially independent designs represent alternative approaches.

Figure 28 illustrates the distribution of finger coverage across the studied hand exoskeletons. The majority of designs aim to assist all five fingers. There is a notable diversity in approaches, with 12.5% of designs focusing on single-finger assistance.

In Figure 29, the distribution of range of motion (ROM) is illustrated. The graph shows data for the metacarpophalangeal (MCP), proximal interphalangeal (PIP), and distal interphalangeal (DIP) joints of each finger. Across all fingers, MCP joints show a preference for ROMs in the 31–90° range, with the highest concentration in the 61–90° category. This aligns with the natural ROM of MCP joints in functional tasks. PIP joints demonstrate the widest range of motion, with significant representations in the 61–90° and 91–120° ranges. DIP joints consistently show smaller ROMs compared to MCP and PIP joints, with most designs in the range of 0–60°.

In Figure 30, the ROM distribution for the thumb joints of the studied exoskeletons is illustrated. The CMC joint, responsible for thumb opposition, shows a preference for smaller ROMs, primarily in the 0–60° range. The thumb MCP joint displays a slightly wider range, with most designs falling in the 31–60° category. The PIP joint shows the widest distribution among thumb joints, with a significant number of designs in the 31–60° and 61–90° ranges.

The total DoF distribution is illustrated in Figure 31. Most designs feature 1–5 DoF, followed by 6–10 DoF, corresponding to designs targeting multiple fingers and more complex movements. The presence of designs with 11–15 DoF and 16+ DoF indicates the development of complex systems.

Figure 32 presents the distribution of safety features implemented in hand exoskeleton designs. Mechanical stops are the most common safety feature, closely followed by inherent compliance. Emergency stops and software limits are frequently used.

The distribution of adaptability features in hand exoskeleton designs is illustrated in Figure 33. Most of the designs are adjustable for different sizes of hands. Customizable components and modular designs contribute to the adaptability criterion.

Box plot analysis illustrated in Figure 34 reveals a systematic relationship between device complexity, functionality, and weight across studied hand exoskeleton designs. The observed weight range, from 25 g to 3700 g, demonstrates significant variation across designs, with median values showing progressive increases from single-finger configurations with a mass of 138 g to five-finger configurations with a mass of 320 g. Single-finger devices exhibit minimal weight variance, whereas five-finger systems display broader distribution patterns, reflecting diverse design approaches from assistive to rehabilitation applications. Mechanical complexity represents a key weight determinant.

The boxplot analysis depicted in Figure 35 demonstrates distinct weight characteristics between soft and rigid/hybrid hand exoskeletons across different finger assistance configurations. Soft exoskeletons maintain consistently low weights, predominantly below 300 g, regardless of assisted finger count, with five-finger designs ranging from 25 g to 300 g. In contrast, rigid/hybrid systems show substantially higher weights and broader distribution patterns, particularly in multi-finger configurations. While soft exoskeletons maintain relatively stable weights across categories, rigid/hybrid devices exhibit marked weight increases with additional finger assistance, suggesting a direct correlation between mechanical complexity and mass.

6. Conclusions

This paper presents a detailed analysis of hand rehabilitation devices based on a review of over 300 systems developed between 2002 and 2024, aiming to reveal the most relevant trends in the development of such devices. The analysis demonstrates a clear trend toward more complex and user-centered designs, with research output peaking between 2016 and 2020. While rigid exoskeletons remain dominant, representing 50% of all designs, there has been substantial growth in soft (35%), and hybrid (15.5%) systems, reflecting increasing attention to user comfort and natural movement patterns and revealing future possible development trends.

In terms of technical implementation, electric actuation remains the predominant choice, representing 55.8% of all systems, with DC motors being the most common actuator type. This might be because of their simpler control systems and their size, which allow these devices to be wearable. The required actuation torque is not high, which is one of the reasons for which the preference for electric actuation is followed by pneumatic systems. Device complexity varies significantly, with 29.4% of designs using five or more actuators while 24.8% employ only one or two actuators, indicating diverse approaches to movement assistance. It also shows that hand rehabilitation can be performed, at least in the initial stages, using underactuated devices, with several fingers performing the same motion together. This approach seems sometimes preferred for practical reasons. Dorsal mechanism placement emerged as the most common configuration, followed by glove-based designs, with most systems implementing independent finger actuation to enable precise control and versatile movements.

Weight analysis demonstrated a clear correlation between functionality and mass. Lighter soft exoskeletons prioritize comfort and mild impairment assistance, while heavier rigid designs offer higher force output and more complex joint control for severe rehabilitation needs.

Analysis of the rehabilitation devices revealed clear relationships between mechanism design choices and clinical applications. Rigid designs usually deliver higher forces that are suitable for severe impairments requiring substantial assistance, while soft systems are suitable for mild to moderate conditions. Hybrid approaches offer moderate force capabilities while maintaining acceptable comfort for medium-term rehabilitation sessions.

Despite these advances, several challenges remain to be addressed in future research. There is a clear need for further investigation into optimal actuation strategies, such as assistive, active-assistive, or resistive ones, targeting different rehabilitation scenarios and the development of standardized evaluation metrics for device performance and therapeutic effectiveness. Integration of advanced sensing and control systems for more personalized rehabilitation approaches represents another crucial area for development.

The field has key challenges that need to be addressed, with missing or incomplete data from source materials being a major concern, a lack of standardized terminology and metrics for direct comparisons, limited long-term clinical studies evaluating therapeutic effectiveness, and the ongoing challenge of developing improved power sources for portable devices. These findings suggest that while significant progress has been made in hand rehabilitation robotics, substantial opportunities remain for improving device functionality, usability, and therapeutic effectiveness.