Design of a Cable-Driven Finger Exoskeleton

Abstract

Motion assistance is a growing need that is linked to the aging of the population, and solutions are required with adequate structures and functionality. Specific attention is paid to finger exoskeletons both for motion exercises and for rehabilitation therapies. The paper presents a new finger exoskeleton structure that is designed with a parallel cable-driven structure with adequate functionality for autonomous use by users. The main advantages can be appreciated in the operational efficiency in assisting finger motion and lightness of the design that also permits easy wearing with the proposed solution installed on a fabric glove. The exoskeleton is designed with a functional characterization through performance analysis thanks to a laboratory prototype that is built for experimental validation tests. The presented prototype appears to be easy to wear and operate with wide possibilities of implementation for motion exercises in rehabilitation and for elderly people.

1. Introduction

Human hands offer a vast range of movement possibilities that most other animals lack. Although individuals with limited operating hands can adapt to perform precise tasks, many daily behaviors still require hand and finger functionality [1]. A hand without finger joints would be significantly less useful, as the independent movement of each finger greatly enhances its capabilities [2]. A human hand has 23 degrees of freedom, with 3 in the wrist and 4 in each finger. These include flexion/extension for each joint of the finger, that is, PIP (Proximal InterPhalangeal), DIP (Distal InterPhalangeal), and MCP (MetaCarpoPhalangeal), and abduction/adduction for the MCP joint. The thumb, with one less phalanx, has similar mobility between the carpal and metacarpal bones (CMC) [3]. The range of finger motion is crucial for daily activities, from grasping objects to complex tasks like typing or tying shoelaces.

This study focuses on finger movements to design a finger exoskeleton that reproduces and assists human finger behavior appropriately. Those finger exoskeletons can aid in medical rehabilitation or can assist with industrial tasks that involve heavy loads or even in daily life. We will specifically analyze the index finger due to its prevalent motion in hand operations.

Research in hand exoskeletons evolves every day, striving towards an ideal design that balances weight, response speed, and ergonomic comfort [4,5]. Different prototypes focus on various aspects, such as the number of degrees of freedom, single-finger functionality, or the use of electromyography (EMG) sensors to detect muscle impulses [6]. Designing hand exoskeletons involves extensive research, considering factors like compatibility, portability, and autonomy. In general, the goal is to restore hand mobility, particularly after strokes, which is profoundly rewarding and has always intrigued scientists. Achieving this requires replicating human hand behavior with a main focus on the degrees of freedom. Not all exoskeletons replicate the MCP joint’s adduction/abduction movement due to system complexity [7,8].

Currently, there are three main methods for achieving hand flexion and extension as follows: rigid structures, flexible components, and pneumatic systems. Rigid structures offer high precision, but they can be bulky and less fluid in movement [9,10]. Flexible designs mimic tendon or muscle action for smoother movements [11]. Exoskeleton systems vary in complexity based on degrees of freedom and number of involved fingers. Single-finger systems simplify design optimization and focus on precision, while multi-finger systems offer precision and force transmission [12,13]. Some systems use EMG to detect muscle impulses and give signals to actuate the exoskeleton, determining the finger to move based on the EMG signal’s position [14].

Other examples of tendon-driven devices are presented in [15,16], highlighting the efficacy of a cable-driven design for assisting devices for rehabilitation. Cable-driven devices for hand rehabilitation are generally limited to flexion or extension only [17], or they are complex and bulky, adopting rigid links as constraints [18,19]. Cables are generally used as transmission elements to actuate the rigid joints of the exoskeletons, to move a motor in the back of the hand or the wrist of a user [20]. However, some light exoskeletons can avoid rigid joints to actuate finger joints by pulling directly on the fingertip only [21]. This design solution maintains most of the functionality while significantly reducing the bulk of the system.

In this paper, a new finger exoskeleton design is proposed based on cable-driven architecture. The proposed design is characterized by a lightweight, low-cost structure that enables self-care at home, improving accessibility when compared to existing rehabilitation devices, which are bulky and expensive. After discussing the biomechanical requirements, the new design is proposed and demonstrated with a 3D-printed prototype to validate its functionality. The paper presents the development of a new light mechanical design for a finger exoskeleton with a prototype for the index finger with the aim of discussing the cable-driven design solution as conceived from careful attention to design and operation requirements in finger motion assistance and from considering previous experience with a linkage-based finger exoskeleton.

2. Materials and Methods

2.1. Requirements

For the design of a new prototype, it is essential to consider requirements on operation and design issues. An exoskeleton that can be adaptable to all users morphologies requires specific characteristics. The size of the different phalanges (proximal, middle, and distal) as well as the flexion and extension angles of the joints (MIP: metacarpal phalanx; PIP: proximal interphalanx; DIP: distal interphalanx) in human finger anatomy are the primary characteristics for designing an exoskeleton finger (Figure 1a). Each joint of the index finger has degrees of freedom for flexion and extension, functioning as a pivot joint (Figure 1b) for the flexion angles α and β. Additionally, the MIP joint has an extra degree of freedom for finger abduction in the abduction angle (Figure 1c). This additional degree of freedom, while present in the human finger, may not be essential for creating an effective and useful exoskeleton. In fact, incorporating this movement complicates the mechanism significantly without providing substantial benefit.

After analyzing different human morphologies, it is possible to identify ranges of values that encompass most of the users physical finger characteristics as summarized in

Table 1. Reference characteristics of the index finger.

	Phalanx Size (Mm)
	Proximal	Middle	Distal
Minimum	30	20	20
Maximum	60	40	40
Average	45	25	25
	Phalanx angle (deg)
	MIP (α, θ)	PIP (β)	DIP (γ)
Range	90, 20	90	90
Minimum	−20, −10	0	0
Maximum	70, 10	90	90

in terms of sizes and motion ranges. The analysis objective will be to develop a new exoskeleton structure that meets as many of the above characteristics as possible. In addition to adhering to the requirements as closely as possible, it is also essential to consider acceptance and practical use by a user. Therefore, practical usability criteria must also be met. Specifically, for user comfort, the system should have relatively low weight, a minimal volume to facilitate easy use, and it should be designed to be comfortable and easy to wear according to reference data, as shown in

Table 2. Lists of reference values of main parameters.

Requirement	Weight (g)	Volume (mm × mm × mm)	Easy Wearing	Comfort
Value	<300 g	<200 × 50 × 30	No assistance to equip and use	Wearable 1 h without discomfort

. The volume values refer to the maximum box shape that can contain the full device worn on an index finger on top of the hand.

2.2. Problems

Most of the exoskeleton systems currently used in medical rehabilitation are custom-designed for each user. This means that a user requiring a finger exoskeleton must wait for a comprehensive study that can develop a system tailored to her/his physical characteristics, such as finger length and mobility range. This process can be time-consuming and costly, whereas hand rehabilitation needs to be conducted as quickly as possible and be accessible to all users. As previously mentioned, an individual who loses or lacks full mobility in her/his fingers is significantly disadvantaged in daily life.

This study aims to address this issue by proposing an adaptable solution for a finger exoskeleton. The paper focuses on the design, development, and testing of a new finger exoskeleton prototype that can be capable of adapting to the morphology of various users. The goal is to provide a practical solution for as many people as possible. The designed system that is presented in this study can be produced in several units of different sizes, significantly reducing the waiting time and cost for users in need of a finger exoskeleton.

2.3. New Design

To design a new prototype, several design ideas can be conceived and considered using CAD software (Inventor 2024). The most interesting design concept that offered the most advantages is the “cable-driven” design. This concept is presented below with explanatory schemes in Figure 2. The cable-driven design concept appears to have the most advantages for meeting the previously outlined requirements. The proposed design consists of six main structural components, each of which is detailed below as summarized in Figure 2a, Figure 3, Figure 4 and Figure 5. Figure 2b shows a CAD design for an artificial finger as a typical structure of a wearable finger exoskeleton that can be used for the conceptual design with its functionality in a motion planning in the finger sagittal plane. The worn rings on the phalanxes are driven by the cable to perform the closing and opening configuration of the finger. The sizes of the rings are determined by hosting the corresponding phalanxes with a cable of an adjustable length according to the assisted finger size. The main components are as follows:

• Support (Figure 3): The main support is a plate where the motors can be fixed in each motor support, and which can be worn on the back of the hand. Therefore, there will be no mobile motor located on a phalanx as in other prototypes. Additionally, there is a small axle to enable the rotation of the first phalanx. There are holes through which the cables pass up to attach directly to each phalanx, allowing suitable cable management, increased efficiency, and reduced encumbrance. Finally, there are four holes underneath the support to attach the handholding, which will be inside the glove worn by the user. The support has a surface area of 70 mm × 90 mm.
• Phalanx 1 (Figure 4a): The first phalanx ring allows the rotation of the proximal phalanx joint. The adjustable holes accommodate a phalanx size ranging from 30 mm to 50 mm with the ring of an inner diameter of 25 mm.
• Phalanx 2 (Figure 4b): The second phalanx ring allows the rotation of the middle phalanx joint. The adjustable holes accommodate a phalanx size ranging from 20 mm to 40 mm with the ring of an inner diameter of 20 mm.
• Phalanx 3 (Figure 4c): The third phalanx ring allows the rotation of the distal phalanx. This phalanx ring can adapt to all sizes over 20 mm with the ring of an inner diameter of 20 mm.
• Pulley and Hand Holding (Figure 5): Additional parts are used to roll cable (Pulley) or also to fix the device to the hand thanks to a glove (Hand holding).
• The sizes have been determined considering an adaptable solution to different users by referring to the data in Table 1.

Figure 3, Figure 4 and Figure 5 give specific insight into the mechanical design with the features of adjustable components of easy wearability with also size indications.

The proposed design in Figure 2, Figure 3, Figure 4 and Figure 5 includes sensors (such as IMU and current sensors) and servomotors with proper installation, as shown in Figure 3 and Figure 4. The finger exoskeleton in Figure 2 is designed to be adapted for all finger sizes from 70 mm to 130 mm of the index, with the possibility to adjust each ring to the corresponding phalanx.

The proposed finger exoskeleton can be modeled as an underactuated mechanism made of a series of parallel modules in series, as illustrated in Figure 6. The first body of the system (0) represents the hand of the user and is defined by the fixed reference frame x₀y₀, centered in A₀. Each ring of the exoskeleton is defined by a frame x_iy_i with origin in its center point A_i (i = {1, 2, 3}, for proximal, medial, and distal phalanx, respectively). A ring is supposed to be worn approximately in the middle of each phalanx, at a measurable distance from the previous and following joint that depends on the user’s biometrics given by lengths l_MC, l_P1, l_P2, l_M2, l_M3, and l_D3, as indicated in Figure 6. The relative position between two successive frames is set by the rotation of the joint that connects them (MCP, PIP, and DIP, respectively), as actuated by the antagonistic pair of cables routed along them, indicated by lengths l_B and l_C. Thus, the transformation matrix between consecutive frames can be written as follows:

$${}^{\mathbf{01}}\mathit{T}=\left[\begin{array}{cccc}\cos {\vartheta }_{MCP}& \sin {\vartheta }_{MCP}& 0& {\mathrm{l}}_{\mathrm{M}\mathrm{C}}+{\mathrm{l}}_{\mathrm{P}1}\cos {\vartheta }_{MCP}\\ \sin {\vartheta }_{MCP}& \cos {\vartheta }_{MCP}& 0& {\mathrm{l}}_{\mathrm{P}1}\sin {\vartheta }_{MCP}\\ 0& 0& 1& 0\\ 0& 0& 0& 1\end{array}\right]$$

(1)

$${}^{\mathbf{12}}\mathit{T}=\left[\begin{array}{cccc}\cos {\vartheta }_{PIP}& \sin {\vartheta }_{PIP}& 0& {\mathrm{l}}_{\mathrm{P}2}+{\mathrm{l}}_{\mathrm{M}2}\cos {\vartheta }_{PIP}\\ \sin {\vartheta }_{PIP}& \cos {\vartheta }_{PIP}& 0& {\mathrm{l}}_{\mathrm{M}2}\sin {\vartheta }_{PIP}\\ 0& 0& 1& 0\\ 0& 0& 0& 1\end{array}\right]$$

(2)

$${}^{\mathbf{23}}\mathit{T}=\left[\begin{array}{cccc}\cos {\vartheta }_{DIP}& \sin {\vartheta }_{DIP}& 0& {\mathrm{l}}_{\mathrm{M}3}+{\mathrm{l}}_{\mathrm{D}3}\cos {\vartheta }_{DIP}\\ \sin {\vartheta }_{DIP}& \cos {\vartheta }_{DIP}& 0& {\mathrm{l}}_{\mathrm{D}3}\sin {\vartheta }_{DIP}\\ 0& 0& 1& 0\\ 0& 0& 0& 1\end{array}\right]$$

(3)

The length of the actuation cables can be written as follows:

$${\mathrm{l}}_{\mathrm{B}}=‖{\mathbf{B}}_{\mathbf{0}}{\mathbf{B}}_{\mathbf{1}}‖+‖{\mathbf{B}}_{\mathbf{1}}{\mathbf{B}}_{\mathbf{2}}‖+‖{\mathbf{B}}_{\mathbf{2}}{\mathbf{B}}_{\mathbf{3}}‖$$

(4)

$${\mathrm{l}}_{\mathrm{C}}=‖{\mathbf{C}}_{\mathbf{0}}{\mathbf{C}}_{\mathbf{1}}‖+‖{\mathbf{C}}_{\mathbf{1}}{\mathbf{C}}_{\mathbf{2}}‖+‖{\mathbf{C}}_{\mathbf{2}}{\mathbf{C}}_{\mathbf{3}}‖$$

(5)

where the vector describing each cable segment can be obtained from module loop-closure equations as follows:

$$‖{\mathbf{B}}_{\mathbf{i}}{\mathbf{B}}_{\mathbf{i}+\mathbf{1}}‖=‖{}^{\mathbf{i}}{\mathbf{B}}_{\mathbf{i}}{\mathbf{A}}_{\mathbf{i}}+{}^{\mathbf{i}}{\mathbf{A}}_{\mathbf{i}}{\mathbf{A}}_{\mathbf{i}+\mathbf{1}}+{}^{\mathbf{i}}{\mathbf{A}}_{\mathbf{i}+1}{\mathbf{B}}_{\mathbf{i}+\mathbf{1}}‖$$

where ${}^{\mathbf{i}}{\mathbf{B}}_{\mathbf{i}}{\mathbf{A}}_{\mathbf{i}}$ is known from geometry;

$${}^{\mathbf{i}}{\mathbf{A}}_{\mathbf{i}}{\mathbf{A}}_{\mathbf{i}+\mathbf{1}}={}^{\mathbf{i},\mathbf{i}+\mathbf{1}}\mathit{T}\left(4, 1:3\right);$$

$${}^{\mathbf{i}}{\mathbf{A}}_{\mathbf{i}+\mathbf{1}}{\mathbf{B}}_{\mathbf{i}+\mathbf{1}}={}^{\mathbf{i},\mathbf{i}+\mathbf{1}}\mathit{T}\left(1:3, 1:3\right){}^{\mathbf{i}}{\mathbf{A}}_{\mathbf{i}+\mathbf{1}}{\mathbf{B}}_{\mathbf{i}+\mathbf{1}}$$

(6)

By integrating Equations (1)–(6), the forward and inverse kinematic problems of the exoskeleton can be modeled in closed-form expressions.

Using CAD software, the movement of the proposed design can be simulated in basic operation modes to test the design and operation feasibility. However, modeling parts like a cable in these software programs can be not so satisfactory. Alternatively, cables can conveniently be replaced by pistons to simulate their pulling actuation action due to actuating cable tension. Thus, the basic functioning of the proposed design has been verified with these pistons simulating the actuation of the pulling cables, and the flexion and extension of the exoskeleton finger mechanism in the planar sagittal configuration can be appreciated, as shown in Figure 7.

The proposed solution is conceived and tested for the motion assistance of the index finger, but its adaptation and extension to the motion assistance of the other fingers of the hand can also be implemented in a configuration that can provide for a multiplicity of similar systems or even more simply by fixing the fingers together with the index finger for a unified operation of all the fingers.

Limitations in the proposed solution can be recognized in the possible dependence of the efficiency of the actuation on the friction between the cables and the pulley and on the deterioration of the cable transmission in prolonged use, in addition to the fact that wearing the rings may be difficult and even not optimal depending on the anatomical conditions of the finger to be assisted. Furthermore, some difficulty can be recognized in ensuring the expected configuration of the cables during operation. To this end, a future development is planned in which both the rings and the cables can be installed in a preassembled configuration on a glove that a user can wear without such difficulties.

2.4. Prototype and Testing Layout

A prototype was assembled using commercial components and the structural parts made by PLA with 3D printing manufacturing by a commercial FlashForge printer (Adventurer 5M printer, FlashForge Zhejiang, China). An easy adaptation to different finger sizes is ensured by properly accommodating the rings in the proper section of phalanxes or by quickly 3D printing the rings with the proper diameter size. The installation of a fabric glove is also helpful for a fairly simple adaptation of the rings and full finger exoskeleton, as shown in Figure 8.

The prototype thus obtained was created for an experimental laboratory campaign to verify its main characteristics and its functionality with a solution that is assembled on a fabric glove to facilitate its wearing. This solution is illustrated in Figure 8a, which highlights the construction aspects of the prototype with its components. The built prototype is for the index finger of the left hand since the pulley on the motor axis for activating the tension of the cables is designed with dimensions to ensure the tension in the cables and the motion assistance action on a sagittal plane of the finger. Figure 8b shows the CAD design of the designed solution on an artificial finger with the reference axes of IMU sensors to monitor the assisted finger motion.

A testing layout has been designed with a specific protocol both in step sequence and respecting ethical issues with human volunteers with user-oriented features to facilitate a volunteer to understand and well perform a test. The prototype operation is monitored during a test by using its own sensors with acquired data that are collected to be stored in a laptop for a postprocessing elaboration and evaluation. A control box is designed to have the electronic components of exoskeleton control and data monitoring in a portable small box containing an Arduino mini controller, a data board with a micro-SD, and a 1000 mAh battery (ensuring operation up to 2 h, as per a potential application of motion assistance).

In particular, the test protocol is run with the following steps:

• Installation of the exoskeleton
  • To wear the glove on the left hand, making sure to position each ring in the middle of the index finger phalanges
  • To position the support so that the right part with the pulley is aligned with the MCP joint.
  • To adjust the cable lengths so that the flexion cable is taut when the finger is bent, and the extension cable is taut when the finger is fully extended.
• Connections to the control box
  • To connect properly the IMUs and motor components to the control box.
• Switch on and run a test
  • To switch on the control box with a green LED lighting up.
  • To run the exoskeleton to move the assisted finger between flexion and extension position.

The designed protocol is run after a volunteer is informed on the test modes and her/his anthropometric data are recorded anonymously. At the end of the test, comments and opinions by the volunteer are also collected as linked to acquired data from the sensors.

3. Results

The built prototype has been used in a testing campaign to check its feasibility and to characterize its performance, looking at features fulfilling main requirements for a low-cost comfort and easy user-oriented operation.

3.1. Testing Model

The feasibility and characterization of the operation performance of the proposed exoskeleton in assisting finger motion are discussed, referring to testing modes in Figure 9 and Figure 10 with test results in Figure 11, Figure 12, Figure 13, Figure 14, Figure 15, Figure 16, Figure 17, Figure 18 and Figure 19 using sensors that equipped the exoskeleton in terms of motion and power consumption. Those results well represent the kinematic features of the exoskeleton mechanism in conjunction with the dynamics driven by the servomotor power and cable tension. The theoretical analysis that is mentioned with the scheme of Figure 6 and Figure 7 is included since those results will not add any specific additional information to being well matched with the test results.

Tests have been designed to check the prototype with a human finger and an artificial finger to have results from different views, such as related to human usage and to functional application. Testing mode consists of motion assistance in the extension–flexion movement of the index finger of the left hand as per the design of the finger exoskeleton with the guiding pulley on the left side of the support.

For tests with the prototype using a human finger, a user wears the glove with the device support that is fixed on its top as shown in Figure 9. Then, the cables are adjusted to the correct length according to the user’s finger anatomy, and finally, they are activated by switching on the device. The pulley size has been sized to allow a 180 deg rotation referring to a complete finger flexion. For a user with a finger approximately 10 cm long, the built prototype fits very well, as shown in Figure 9. This satisfactory result in full extension that is followed by complete flexion of the assisted finger is illustrated in the snapshot of Figure 9.

Figure 9. Snapshot of a test with the prototype of a finger exoskeleton in assisted extension–flexion motion of a human finger: (a) intial configuration; (b) intermediate configuration; (c) final configuration.

Figure 9. Snapshot of a test with the prototype of a finger exoskeleton in assisted extension–flexion motion of a human finger: (a) intial configuration; (b) intermediate configuration; (c) final configuration.

For tests with an artificial finger, the device support and pulleys have been resized and adapted to a 1.5 scale as shown in Figure 10. Additionally, the motor is selected differently for larger sizes. The testing follows the same protocol as for the prototype with a human index finger in Figure 9. The experienced behavior shows very good results with the artificial finger’s movement closely reproducing the flexion of a healthy human index finger, as illustrated in the snapshot of Figure 10, that is comparable with that one in Figure 9.

Figure 10. Snapshot of a test with the prototype of a finger exoskeleton in assisted extension–flexion motion of an artificial finger. (a) intial configuration; (b) intermediate configuration; (c) final configuration.

Figure 10. Snapshot of a test with the prototype of a finger exoskeleton in assisted extension–flexion motion of an artificial finger. (a) intial configuration; (b) intermediate configuration; (c) final configuration.

3.2. Test Results

During the motion tests conducted with the two solutions presented above in Figure 9 and Figure 10, various sensors were positioned on the finger exoskeleton mechanism to monitor the exoskeleton behavior accurately. Like in the previous prototype ExoFinger [6,10], two IMUs were placed on the finger rings as follows: one on the proximal phalanx and the other on the distal phalanx. Additionally, a current sensor was connected in series with the motor to monitor the power consumption. These sensors continuously provide six data sets as follows: angular velocity around the X, Y, and Z axes of the IMU reference and linear acceleration along these axes yet. The angular velocity helps to identify when the finger starts moving, but it does not provide insightful information about the finger’s behavior and movement. But linear acceleration offers the most valuable data since it refers to the kinematics of the assisted motion.

Test results with a human finger, as in Figure 9, are reported in Figure 11, Figure 12, Figure 13, Figure 14 and Figure 15, and those with an artificial finger, as in Figure 10, are shown in Figure 16, Figure 17, Figure 18 and Figure 19.

Referring to a test with a human finger, in Figure 11a, the assisted motion is detected in angular velocity with consistent numerical noise, very likely due to the tremor functioning of the assisted finger motion following the not sufficiently smooth cable actuation. However, it can be noted a cyclic acquisition of the data corresponding to the flexion–extension of the assisted finger with primary motion in the sagittal plane indicated by angular velocities around X and Y IMU axes. In Figure 11b, during the first two seconds of the test, the linear acceleration fluctuates rapidly over short periods, corresponding to the motor’s initialization phase since it automatically positions itself at the starting point. Subsequently, the acceleration follows a sinusoidal pattern with a period of about 4 s, especially along Y and Z axes, which are the primary axes of movement since the assisted finger moves in the finger sagittal plane. The motion is well detected with Y acceleration as referring to the main motion component showing the three cycles on flexion–extension of a test. The acceleration along the X-axis with nonzero values indicates that the finger does not move with a planar movement in the sagittal plane, but it is slightly twisted. The acquired sinusoidal pattern clearly illustrates the finger’s flexion and extension movements along the Y-axis as follows: when the acceleration is positive, the finger is in a descending flexion phase, and when it is negative, the finger is in an ascending extension phase.

Figure 11. Results acquired by IMU 1 in Figure 8 from a test with a human finger in Figure 9: (a) angular velocity and (b) acceleration components.

Figure 11. Results acquired by IMU 1 in Figure 8 from a test with a human finger in Figure 9: (a) angular velocity and (b) acceleration components.

Considering the IMU axis reference in Figure 8b, the magnitude of the linear acceleration can be calculated free of gravity by A = (Ax² + Ay² + Az²)^0.5 − g (being g the gravity acceleration), with results in Figure 12. The computed values are limited within a range of ±1 m/s², indicating, in general, small values of a smooth assisted motion but with a few sudden changes, as reported by acceleration peaks corresponding mainly to the inversion of the motion from flexion to extension of the assisted finger.

Figure 12. Computed acceleration magnitude from results acquired by IMU 1 in Figure 8 during a test with a human finger in Figure 9.

Figure 12. Computed acceleration magnitude from results acquired by IMU 1 in Figure 8 during a test with a human finger in Figure 9.

Similarly, the acquired values are from IMU 2 on the middle phalanx, Figure 8, but with some different results, mainly in angular velocity. For the proximal phalanx, the movement is detected as a 60° rotation around the MIP joint with more evident characteristics both in values and cyclic displacements, being the main phalanx involved in the flexion–extension motion. Significant rotation and therefore angular velocity are around the X-axis of IMU2, as shown in Figure 13a. Regarding the linear acceleration of the distal phalanx in Figure 13b, the large values along the X axis indicate a slight misalignment from the sagittal plane as per the not-fully-correct wearing of the distal ring as shown in Figure 9 since the IMU2 house is slightly rotated. Nevertheless, the behavior of the flexion–extension motion is still detected with its cyclic timing, likewise the values along the Y axis, but in counter phase. The values along the Y-axis illustrate the finger moving forward and backward horizontally during the flexion–extension motion. Along the Z-axis, the values refer to the finger distal phalanx moving up and down vertically with rather small acceleration values. When the Ay value is at its minimum, it corresponds to the maximum of the Az values, indicating that the finger is extended and raised towards its horizontal position. These acquired data of the acceleration components, as shown in Figure 13b, well characterize the movement of the distal phalanx of the finger during the tested finger flexion–extension motion.

Figure 13. Results acquired by IMU 2 in Figure 8 from a test with a human finger in Figure 9: (a) angular velocity and (b) acceleration components.

Figure 13. Results acquired by IMU 2 in Figure 8 from a test with a human finger in Figure 9: (a) angular velocity and (b) acceleration components.

Similar to the acquired IMU 1 acceleration data, the magnitude of acceleration of the distal phalanx from IMU2 can be calculated as shown in Figure 14 using the formula for free gravity computation. In the plot, peaks with significant variation over the average range within ±1 m/s² occur around 4 s, 7 s, and 12 s when Az reaches its maximum, indicating that the finger is extended horizontally when starting the inversion of the motion during the cyclic test exercise.

Figure 14. Computed acceleration magnitude from results acquired by IMU 2 in Figure 7 during a test with a human finger in Figure 8.

Figure 14. Computed acceleration magnitude from results acquired by IMU 2 in Figure 7 during a test with a human finger in Figure 8.

In Figure 15, data are reported as acquired by the current sensor that is connected in series with the motor to measure the motor’s current consumption when powered by a 5V supply. The results show a variation between 420 mA and 500 mA, which is a consistent energy consumption for the used small motor, indicating active assistance on the finger motion. This level of power consumption ensures sufficient torque, enabling the motor to transmit a significant force through the cables to guide the assisted finger during the prescribed cyclic motion during the test in Figure 9.

Figure 15. Results acquired by IMU 1 in Figure 8 as current consumption from a test with a human finger in Figure 9.

Figure 15. Results acquired by IMU 1 in Figure 8 as current consumption from a test with a human finger in Figure 9.

The same tests on the human finger as in Figure 9 were conducted with the artificial finger as in Figure 10, obtaining comparable results, as shown in Figure 16, Figure 17, Figure 18 and Figure 19. The data from IMU 1, placed on the proximal phalanx, are shown in Figure 15, referring to angular velocity, and in Figure 16b, referring to linear acceleration. The angular velocity is detected with a time history very similar to that acquired for the human finger in Figure 13a. It is to note that the finger’s rotation occurs around the X-axis, with detected values within ±12°/s that are smaller than in the case with the human finger, while the other components are measured with similar values. This can indicate that there is no additional motion due to the finger structure due to finger action, as it is indeed detected also with the acquired acceleration reported in Figure 16b. The finger movement occurs in the YZ plane and effectively illustrates the rotation around the MCP joint with Ay and Az values referring to the cyclic finger flexion–extension motion during the simulated test.

The motion is well detected with Y acceleration as referring to the main motion component showing the three cycles on flexion–extension of a test. The acceleration along the X-axis is also obtained with nonzero values, indicating that the finger does not move with a planar movement in the finger sagittal plane, but it is slightly twisted as also probably due to the not-exact location of the IMU sensor in a horizontal posture. The obtained cyclic time evolution of the components Ay and Az with similar values with respect to the test with the human finger clearly illustrates the flexion and extension movements with some numerical noise and small peaks very likely due to backlash and friction in the wood finger model.

Figure 16. Results acquired by IMU 1 in Figure 8 from a test with the artificial finger in Figure 10: (a) angular velocity and (b) acceleration components.

Figure 16. Results acquired by IMU 1 in Figure 8 from a test with the artificial finger in Figure 10: (a) angular velocity and (b) acceleration components.

The magnitude of the linear acceleration is calculated free of gravity, as shown in Figure 17, with values that are limited within a range of ±2 m/s² with small values but a little larger than in the test with a human finger. This indicates smooth assisted motion but with a few sudden changes as acceleration peaks, corresponding mainly to the inversion of the motion from flexion to extension of the assisted finger.

The acquired values from IMU 2 of the angular velocity of the middle phalanx are reported in Figure 18a with some small differences in values and time evolution with respect to those from tests with the human finger as probably due to both the active participation of the finger during human tests and model limitations in the finger design and operation. The major values are detected with respect to the X axis within a range of ±40°/s, confirming the operation in the sagittal plane with small error movements with the respect to X axis. The reported peaks are produced by unexpected lateral motion of the artificial finger due to its mechanical design with free revolute joints among the phalanx bodies. Still evident are characteristics of cyclic displacements since the distal phalanx is the one mainly involved in the flexion–extension motion.

Figure 17. Results acquired by IMU 1 in Figure 8 as acceleration magnitudes from a test with an artificial finger in Figure 10.

Figure 17. Results acquired by IMU 1 in Figure 8 as acceleration magnitudes from a test with an artificial finger in Figure 10.

In Figure 18b, the linear acceleration of the distal phalanx is reported with the large cyclic values along the Z axis in a ±10 m/s² range with a time history of Ay along the Y axis in a 0 to 10 m/s² range, well representing the flexion–extension motion of the finger imposed in the simulated test. While Ay is computed with similar values to those of the human test, the values of Az show differences with smaller values in the area of negative values, indicating very likely an active participation of the human finger in a test differently from what is simulated with the artificial finger. The very small values of Ax along the X axis indicate a satisfactory motion in a sagittal plane as already detected in the computed results for the angular velocity. These acquired data of the acceleration components, as shown in Figure 18b, well characterize the assisted movement of the finger distal phalanx during the simulated finger flexion–extension motion.

Figure 18. Results acquired by IMU 2 in Figure 8 from a test with an artificial finger in Figure 10: (a) angular velocity and (b) acceleration components.

Figure 18. Results acquired by IMU 2 in Figure 8 from a test with an artificial finger in Figure 10: (a) angular velocity and (b) acceleration components.

In Figure 19, the current consumption of the motor for the artificial finger is computed close to 500 mA also with a time history like in the case of human finger tests. This current is relatively low for the motor of the selected size in the simulation, but it is larger than the one used in the prototype with the human finger.

Figure 19. Results acquired by IMU 1 in Figure 8 as current consumption from a test with an artificial finger in Figure 10.

Figure 19. Results acquired by IMU 1 in Figure 8 as current consumption from a test with an artificial finger in Figure 10.

The results of the simulated operation with the artificial finger are very similar to those obtained with the tests with the human finger, indicating that the tested operation well represents the operation of the designed cable-driven finger exoskeleton in assisting a human finger. The most significant difference between the two reported tests is that the values are larger in the case of the artificial finger. This is expected since the artificial finger is 1.5 times larger than the human finger, leading to correspondingly larger values for both angular velocity and linear acceleration. Thus, the prototype is unlikely to cause any issues for a user as the assisted movements it produces are not harmful. The tested movements required of the user are comparable to those of a human finger in daily activities and are therefore not traumatic.

The simulated tests with the artificial finger, whose results have confirmed the results from laboratory tests with human fingers, have clarified the successful assistance of the designed cable-driven finger exoskeleton with reasonably satisfactory finger-assisted motion since the artificial finger simulates the operation with no active or reactive finger as it could occur with the human fingers of the volunteers. This ultimately indicates that the designed cable-driven finger exoskeleton is unlikely to cause any risk for a user since the assisted motion is not harmful.

The tests reported with their characterization through the acquired characteristics are examples that are selected from those carried out with tests repeated at least three times each. The purpose of these tests is the validation of the feasibility of the proposed solution and its functional characterization that meets the expectations of motion assistance for fingers. A future development of the study is planned with an experimental campaign with an adequately high number of subjects and test modes that can definitively characterize the suitability of the proposed exoskeleton for physiotherapy and motion exercise applications, also considering long-term efficiency issues.

4. Discussion

The innovative aspects of the proposed cable-driven solution can be recognized in the modular structure with cable actuation, which differs from the current exoskeletons available on the market and proposed in the literature, as reported in the cited references. Those innovation aspects have been evaluated for a patent request under examination considering the main features in the efficient, easy-to-wear operation of a portable light comfort design. The existing solutions are characterized by solutions with articulated mechanisms or cables with muscular-type solutions with bulky mechanical designs, while the proposed solution consists of three light, wearable rings with a small platform. The cable actuation is designed with cables that are actuated by a small actuator that can be installed on a light platform on the back of the hand in proximity to the assisted index finger. The technical aspects of the proposed design are detailed in both kinematic and mechanical designs, also through CAD design models showing the components and structural elements with dimensional details in Figure 2, Figure 3, Figure 4, Figure 5, Figure 6, Figure 7 and Figure 8. Main features of the proposed design can be recognized in terms of light design, wearability, motion assistance efficiency, and user-oriented operation with different and better characteristics with respect to the existing solution, whose illustrative survey refers to the references [1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21].

Functionality-wise, many previous hand exoskeletons are based on rigid mechanisms, which impose a significant load on the patients and cannot adapt to hand structure, such as, for example, in the solutions reported in [1,6,7,8,9,10,11,12,19,20,21,22]. A few lightweight, cable-driven exoskeletons are proposed as soft glove-based solutions, solving the issues of rigid-link exoskeletons, as in [4,17,18,23]. However, they still face some of the same challenges, such as both rigid-link and glove-based mechanisms being strongly influenced by hand size and biometrics. Furthermore, the lack of defined fixture points makes glove-based designs less effective in their assistive action.

The proposed design is placed at the intersection of these solutions, mitigating the limitations of the above limits. The ring-based design can be cheaply manufactured for a large variety of sizes and can be used to reliably, robustly fix the exoskeleton to the wearer’s body, even with or without a glove installation. Its overall weight of less than 50 gr ensures easy portability and a low-cost solution. The rigid routing points allow for high efficiency, but, at the same time, the lack of rigid joints avoids high and potentially harmful reaction forces on the wearer’s hand so that it ensures convenient comfort in usage. This also allows the exoskeleton to adapt naturally to any hand size, a feature that neither glove-based nor rigid-link-based solutions could achieve.

From a theoretical perspective, the proposed cable-driven solution manages to actuate each joint of the human hand with a single degree of freedom, underactuating each finger by leveraging the natural coupling between its phalanges, that is, by pulling the distal phalanx, all the other phalanges naturally bend with it. The cables thus act like an agonist/antagonist muscle, rather than forcing motion through an external linkage or rigid mechanism, replicating the functionality of the human body rather than constraining it. The experimental characterization shows how this mechanism can guide a bending of approximately 150 deg, with smooth behavior without any sudden acceleration or change in motion characteristics.

5. Conclusions

This paper presents results for the design and testing of a new cable-driven exoskeleton for the index finger with main features of being easily adaptable to different finger sizes, ease of use, and no need for external assistance for its operation. Furthermore, the use of cables allows for a compact overall design when compared to existing finger exoskeleton devices. Additionally, the operation of these cables facilitates very smooth movement and enhances user comfort. The proposed design is built with a prototype for a left finger since the size of the pulley guiding the flexion cable beneath the finger is proper tension. The reported test results show satisfactory motion assistance with the main targeted features in portability, easy wearing, comfort, user-independent usage, and monitoring main operation characteristics. The current prototype allows for length adjustment but not for width adjustment to user finger anatomy so that future improvement is planned to size the phalanx rings to accommodate different finger widths. The paper’s contribution can be recognized in presenting a novel light design of a cable-driven finger exoskeleton with a modular design with small wearable rings with a sensorization on board of the rings yet. Its operation performance has been tested successfully in assisting finger motion exercises with features of usage by users autonomously in a home environment with small power consumption.

6. Patents

Ceccarelli, M.; Russo, M.; Vaisson, T., Cable-driven exoskeleton for index finger. Italy Patent request No. 102024000019144, 21 August 2024.