Development, Research, Optimization and Experiment of Exoskeleton Robot for Hand Rehabilitation Training

Abstract

As one of the most influential symptoms of daily life after stroke, hand dysfunction has seriously affected the quality of life of patients and families. At present, the commonly used rehabilitation method is to carry out continuous passive training on the patient’s fingers with the help of physical therapists, so as to promote the rehabilitation of the hands. However, this kind of therapist-assisted rehabilitation greatly increases the cost of rehabilitation treatment and is not conducive to the promotion of household use. Many exoskeleton hand rehabilitation robots still lack in overall weight and control, resulting in the application potential of exoskeleton hand rehabilitation robots not being fully developed, and the commercial and clinical success cases are limited. The research of this paper focuses on the structural design and electronic control design of the exoskeleton hand rehabilitation robot. Through the design of an exoskeleton hand rehabilitation robot suitable for human hands, the kinematics parameters are obtained by kinematics simulation, and the lightweight design of the hand rehabilitation robot is completed by using topology optimization. At the same time, this paper shows the development technology of the rehabilitation robot control system. We hope that through the subsequent product development, the exoskeleton hand rehabilitation robot studied in this paper can be applied in the future.

1. Introduction

The hand is one of the most complex motion and sensory organs of human body. A single hand has 24 degrees of freedom, and the finger part alone has 21 degrees of freedom. Therefore, the movement pattern of the finger is very complex and can complete many delicate movements [1,2].

However, the function of the human hand will deteriorate due to bone diseases, hand injuries, nerve injuries and other reasons. Among many diseases, stroke can cause a variety of neurological damage to patients, and it is a major factor leading to the impairment of hand function [3,4,5].

In recent years, with the continuous development of medical technology, the death rate of stroke has gradually decreased. However, stroke still has a very high disability rate and has a great impact on life. Among the survivors of stroke, about 80~90% of the patients will leave limb motor function defects and lose the ability to live independently and exercise [6,7].

In China, the number of new stroke patients is more than 2 million every year. With the change of social and demographic structure in China, the aging problem is further aggravated. It is expected that the number of stroke patients will rise further in the long future. Motor dysfunction after stroke has become one of the hotspots of modern rehabilitation medicine [8,9].

For the motor dysfunction caused by stroke, the clinical rehabilitation treatment is mainly to remodel the damaged nerve. Generally, the therapist exercises the patient’s limbs and muscles to promote the body’s learning and improve the damaged nerves through repeated movements of the limbs. Traditionally, patients go to rehabilitation institutions to assist rehabilitation training through professional physiotherapists to assist patients to complete a series of physiotherapy exercises to achieve the goal of functional rehabilitation. However, this method is not very effective. Many patients cannot rely on rehabilitation training because of its high cost and time-consuming labor [10].

To solve this problem, the rehabilitation robot has many advantages, such as repetitive work, not limited by time and place. Therefore, the rehabilitation robot that assists limb movement rehabilitation has become a research hotspot at home and abroad.

The hand functional rehabilitation robot is an important limb rehabilitation device. Its main task is to assist the hand joints and fingers to complete the rehabilitation actions of bending and stretching. Additionally, the patient can complete the continuous passive exercise therapy provided by the doctor.

Research on the hand exoskeleton began in the 1990s and the hand exoskeleton developed from the industrial manipulator.

Nicola Secciani et al. (Department of Industrial Engineering, University of Florence) developed portable, wearable and highly customizable hand exoskeletons to assist patients with hand disabilities, and defines a patient-centered design strategy to tailor the device to the needs of different users [11]. Lightweight underactuated RACA neurorehabilitation exoskeleton designed and characterized by Victor Moreno-SanJuan (Institute of Advanced Production Technology, University of Valladolid) [12]. Mine Sarac (Frisoli PERCRO Lab) proposed an underactuated hand exoskeleton that adapts to the shape and size of objects during grasping tasks [13].

A two-finger exoskeleton with force feedback for the teleoperation of mobile robots was designed by Manuel Arias-Montiel et al. (Institute of Electronics and Mechatronics, Mistka University of Technology) for the human hand, and it does not limit the range of motion of the human hand, nor does it have any mechanical interference between the exoskeleton and the user’s fingers [14].

Eric M. Refour (Virginia Tech Department of Mechanical Engineering) proposes a novel, thin, and lightweight hand exoskeleton linkage with force control enabled by compact tandem elastic actuators [15]. Giuseppe Carbone (University of Calabria) presents a systematic design approach for a novel dual-DOF actuated linkage for motion-assisted finger exoskeletons [16]. Marco Ceccarelli (Laboratory of Robotic Mechatronics, Department of Industrial Engineering, University of Tovergata) describes a finger exoskeleton with a prototype of the ExoFinger, the experiments describe the proper operation of the ExoFinger. In addition, the skeleton has various functions for motion control and medical monitoring sensors [17].

Nuevo Leon developed a wearable finger exoskeleton, motion transmission is based on a linkage mechanism that allows the possibility of coupled phalanx movements, thus using only one active degree of freedom, one linear for each finger drive, meaning that you can achieve the natural movement of the hand [18]. Husam Almusawi (Faculty of Information, University of Debrecen) designed and developed a novel cost-effective mechatronic system for continuous passive motion training of fingers and wrists [19]. Anirban Chowdhury (Department of Mechanical Engineering, Indian Institute of Technology) proposed a stroke rehabilitation hand exoskeleton with flexible structure and control, and verified its potential clinical effectiveness through preliminary clinical trials in stroke patients [20].

Uikyum Kim et al. (Department of Mechanical Engineering, Asia University) developed a dexterous anthropomorphic robotic hand with integrated linkage actuation. The mechanism is constructed through a fusion of parallel and series mechanisms to achieve two-DOF motion in the metacarpophalangeal joint and single-DOF motion in the proximal interphalangeal joint through a link combination [21]. Teja Vanteddu (Department of Mechanical Engineering at Virginia Tech) designed a robotic exoskeleton glove for stable grip control [22]. Jenny Carolina Castiblanco (University of Pontificia) has solved the problem of how to provide active support through a robot-assisted exoskeleton, by developing a new closed-loop structure that continuously measures electromyographic signals (EMG) to adjust the assistance provided by the exoskeleton [23].

Ouyang Kun (Southern Taiwan University of Science and Technology) designed an exoskeleton assistive device to help hemiplegic stroke patients stretch their fingers and open their palms to simulate the effect of rehabilitation [24]. Inseong Jo (Department of Mechanical Engineering) developed a portable and spring-guided hand exoskeleton system for exercising finger flexion and extension [25].

Zhang Fuhai from the State Key Laboratory of Robotics and Systems, Harbin Institute of Technology, designed an active and passive control system for the exoskeleton rehabilitation hand, which can adapt to fingers of different thicknesses and lengths to prevent secondary injuries [26]. Li Min (School of Mechanical Engineering, Xi’an Jiaotong University) designed an attention-controlled finger extension and flexion rehabilitation hand based on a rigid-flexible combined mechanism. Due to the design and characteristics of its new multi-segment mechanism driven by steel springs, it can help the fingers to stretch and curve [27]. Cheng Chang (Institute of Automation, Chinese Academy of Sciences) proposed a novel controller for finger position detection and tracking control of a wearable hand rehabilitation robot [28]. Li Ke (Laboratory of Rehabilitation Engineering, School of Control Science and Engineering, Shandong University) designed a wearable robotic hand exoskeleton based on surface EMG signals, which has greater freedom of movement, greater range of joint motion, and can be freely controlled according to motion intentions [29]. Zhiyuan Lu (Department of Physical Medicine and Rehabilitation, University of Texas Health Science Center at Houston) developed an intent-driven hand training system using sEMG as intention recognition device [30].

The hand functional rehabilitation training devices and exoskeleton devices developed by different research institutions have different structures and driving modes. The main structure of a hand rehabilitation robot includes linkage mechanism, micro cylinder mechanism, cable drive mechanism, gear mechanism, flexible mechanism, and so on.

Through mechanical design, the hand rehabilitation robot can achieve the synchronous hand movement requirements, thus achieving the rehabilitation training effect. The motion accuracy, bionics and portability of the structure are particularly important.

2. Materials and Methods

2.1. Design of Hand Function Rehabilitation Robot

2.1.1. Hand Structure and Structure Design of Rehabilitation Robot

Hand dysfunction is one of the most common dysfunctions in stroke patients. It is one of the main factors hindering patients’ normal life, work and social life, as well as returning to family and society, causing a heavy burden on patients and families. The function of the hand is complex, including many functions such as movement and sensation, and is also closely related to the advanced functions of the brain such as cognition. Its movement has the characteristics of a large range of activities and complex and diverse forms of movement. It is the most important effector organ of purposeful and instrumental movement [31,32].

Of the 206 bones in the human body, 54 are in the hand, equivalent to a quarter of the total number of human bones; The muscle structures that drive them are also extremely complex. At the same time, the joints of fingers have more degrees of freedom and various motion functions. In order to use robots to realize these functions, many dexterous anthropomorphic robot hands have been developed. In order to perform an effective grasping action, many effective manipulators with adaptive grasping or low degree of freedom (DOF) have been developed [33,34,35].

In different existing architectures, we decided to adopt the exoskeleton based on connecting rod coupling linkage. The connecting rod driving mechanisms have inherent advantages, such as ruggedness, easy manufacture and maintenance; however, they also have disadvantages, such as the difficulty in carrying out multi-degree of freedom motions in a small space while maintaining sufficient workspace. Therefore, it is of great significance to realize a linkage-driven robot finger mechanism with human-like finger motions.

The motion transmission is based on the linkage mechanism to realize the coupling motion of each joint of the finger. The rod can transmit tensile and compressive loads. Therefore, each finger can only use one degree of freedom, that is, it can perform the extension (the most needed motion in rehabilitation) and flexion (auxiliary) motion of the finger.

Specifically, the designed exoskeleton rehabilitation robot is mainly composed of four identical finger modules, which drive the index finger, middle finger, ring finger and little finger, respectively. Each finger module has a degree of freedom actively driven by a linear driver.

The structural design drawing is shown in Figure 1. The linear driver is constrained on the fixed frame and connected with the proximal phalanx fixed block of the finger. When the driver moves in a straight line, it drives the proximal phalanx fixed block of the finger to move around the circular guide rail with the shaft, and then drives the connecting rod to move. The load transmitted by the connecting rod drives the middle phalanx fixed block to move along the circular guide rail, so as to drive the middle phalanx fixed block to move, realizing the bending action of the finger. Similarly, the linear driver moves in the opposite direction to complete the extension movement of the finger.

2.1.2. Design of Control Hardware

The control system of hand functional rehabilitation robot is mainly com-posed of main chip, power supply, WIFI, touch screen, actuator control module, etc. The control system is shown in Figure 2.

We used linear actuators produced in a Canadian company called Actuonix Motion Devices, as shown in Figure 3, which has an onboard software-based digital microcontroller. Although this actuator we used in the hand robot contains a built-in control board, the control board is used to receive control signals and feedback position, the microcontroller is not user-programmable.

We use the four pins in the signal interface of the linear actuator for control and feedback. They are ground terminal, power supply terminal, control signal input terminal, and position feedback signal terminal.

Two of the pins GND and +6 V are used to supply power to the linear actuator. The control signal input pin can receive a 0–5 V voltage signal or a 5 V PWM signal to control the stroke of the linear actuator from 0–100%. We used a 5 Volt 1 kHz square wave on actuator control. In this case, the duty cycle of the control input is the same as the stroke ratio. The position feedback signal pin linearly outputs a voltage signal of 0–3.3 V, according to the actual position ratio of the linear actuator.

The drive interface of each linear actuator uses the four-pin interface of xh254. The interface includes grounding, 6 V power supply, linear actuator control terminal and linear motor position detection.

The power supply of the screen in the system is 5 V, the driving voltage of the linear actuator is 6 V, and the voltage of the single chip microcomputer system is 3.3 V. Therefore, in this paper, the hardware system selects 7.4 V power supply and provides different voltages through different DC–DC and linear converters step-down systems. Figure 4 shows the power supply and DC–DC step-down system. Different power chips are used to realize the power supply of 3.3 V, 5 V and 6 V. At the same time, because the PWM control signal voltage provided by the single chip microcomputer is 3.3 V, which is inconsistent with the driving requirements of the linear actuator, sn74lvc4245a of “octal bus translator and 3.3 V to 5 V shifter with 3-state outputs” is used to regulate the PWM control signal in this paper.

Stm32f405 chip is used for the main control of the hand function robot, as shown in Figure 5.

Figure 4 and Figure 5 show the hardware design of the rehabilitation robot. In the schematic part, the placement direction or labeling of some devices is not standardized.

Figure 6 shows the main control circuit board of the exoskeleton hand rehabilitation robot. Xh2 is used for motor control interface, UART screen interface, UART debugging interface and battery interface. MX1 is used for program debugging and sensor interface.

2.1.3. Rehabilitation Training Model and Software

Four modes are designed for the hand functional rehabilitation robot, including finger training, assistance training, grasping training and stimulation mode. In the daily rehabilitation training, the most important ones are finger training and grip training. Our software interface is shown in Figure 7.

In the software interface as shown in Figure 7, there are four training modes: pinch two fingers, strength assistance, grasping training, and stimulus mode.

The control flow chart of the hardware system is shown in Figure 8.

2.2. Kinematics and Motion Simulation of Hand Exoskeleton Robot

The simplified model of the hand functional rehabilitation robot designed in this paper is shown in Figure 9, in which the MCP (metacarpophalangeal) and PIP (proximal interphalangeal) joints of the fingers are combined and linked through the connecting rod structure.

The mechanism model of finger part was simplified and the model was built in AD-AMS software, as shown in Figure 10.

2.3. Optimal Design of Exoskeleton Hand Robot Based on Topology Optimization

2.3.1. Finite Element Analysis of Hand Rehabilitation Robot

In order to reduce weight, reduce material waste and enhance durability, topology optimization is carried out for the design structure. The model is imported into ANSYS Workbench and connected to the static structural module. The material attribute is defined as aluminum alloy, and the model weight is 0.45 kg. Firstly, the mesh is divided, the constraints and loads are defined, and the maximum deformation and maximum stress values are solved, which are 0.25468 and 93.878 mpa, respectively. Then, it was connected to the topology optimization module and the optimization target under the same constraints and loads was selected.

The finite element analysis process is shown in Figure 11. Figure 11a is the schematic diagram of the structure after importing ANSYS. Figure 11b is the model after meshing, we used tetrahedron mesh to generate 113,741 nodes and 395,150 elements. Figure 11c is the stress analysis model and Figure 11d is the strain model.

2.3.2. Robot Optimization Based on Topology Method

Select the arm, wrist and finger support as the optimization target, set the optimization quality as 40%, select the optimization area and start topology optimization. Output the geometry file after topology optimization, edit the file and convert it into solid geometry model, reassemble it into the assembly, and import it into static structural again. The model weight is 0.39 kg. The maximum deformation and maximum stress are 1.0278 and 78.696 mpa, respectively.

As shown in Figure 12, the topology optimization results are shown, and Figure 12a is the topology optimization generation model. Since the topology optimization-generated model is irregular, the optimized model is obtained by repairing it with 3D modeling software, as shown in Figure 12b. For the optimized structure, conduct the finite element analysis again, as shown in Figure 12c for the stress model and Figure 12d for the strain model.

The boundary conditions for the FE model in this paper is shown in Figure 13.

3. Results

3.1. Hand Function Rehabilitation Robot Prototype

Through the structure design, hardware design and software system development described in this paper, combined with 3D printing, CNC processing, circuit processing and programming, we have completed the complete machine of the hand rehabilitation robot. The prototype of exoskeleton type hand functional rehabilitation robot is shown in Figure 14.

3.2. Hand Rehabilitation Prototype and Parameter Comparison after Topology Optimization

Through the mechanical model simplification and model establishment described above, the whole machine simulation model of the hand rehabilitation robot is completed. In the simulation software, the same drive as the real electric cylinder is applied to the model. Driving the finger to complete the desired motion. By collecting the motion data of fingertips and MCP joints, the motion trajectories of different fingers were drawn. The finger root was set as the motion origin, and the motion range of MCP and fingertip of different fingers are shown in Figure 15.

3.3. Hand Rehabilitation Prototype and Parameter Comparison after Topology Optimization

We consider the lightweight and weight-reduction design of the whole machine, and optimize the components through the topology optimization mentioned above. Finally, the weight-reduction version of the hand rehabilitation robot is designed and the whole machine is manufactured. The hand rehabilitation robot after topology optimization is shown in Figure 16.

Figure 17 shows the overall mass, strain information and stress information of the whole machine before and after weight reduction optimization.

It can be seen from the simulation results that the quality of the whole machine has been reduced to a certain extent under the condition of meeting the application requirements, and the feasibility of the structure has been ensured.

4. Discussion

The exoskeleton hand rehabilitation robot designed in this paper achieves good man-machine matching through kinematics design. This paper presents the design of the core driver, control system and software interface used in the design of the exoskeleton hand rehabilitation robot, and realizes the intelligent human–machine control of the exoskeleton hand rehabilitation robot. At the same time, the kinematics of the exoskeleton hand rehabilitation robot is analyzed by simulation software, and the weight reduction in the whole hand rehabilitation robot is realized by topology optimization.

The hand rehabilitation robot designed in this paper has a certain function of continuous passive rehabilitation, but the overall appearance is still too industrialized, and the appearance design lacks human elements. In the follow-up study, we plan to conduct in-depth development work for the product.

In addition to the appearance, a very important factor in the commercialization of the hand rehabilitation robot is the cost. However, the rehabilitation robot designed in this paper uses a commercial servo actuator. This basically uses half the cost of the whole hand rehabilitation robot. We are also exploring and developing alternative linear actuators. It is expected that through the development of core execution devices, the cost can be reduced, so as to make hand-held rehabilitation robot products.

5. Conclusions

In this work, we developed an exoskeleton robot for hand functional rehabilitation according to the actual needs of hand functional rehabilitation. We designed the finger parts of the exoskeleton robot through the modular design idea. The frame is used to combine the finger parts and form the overall mechanical structure of the exoskeleton hand rehabilitation robot. Subsequently, the core circuit suitable for linear servo drive is designed. At the same time, this paper develops the control algorithm and UI control interface suitable for hand function rehabilitation, thus realizing the intelligent control of hand rehabilitation robot. Considering the overall weight requirement of the exoskeleton hand rehabilitation robot, this paper uses finite element analysis and topology optimization. On the premise of ensuring the effectiveness and stability of the whole structure, the optimization and weight reduction in the whole structure are realized.

This paper shows the technologies involved in the hand rehabilitation robot we developed, including the technical details of mechanical design, hardware circuit design and program development, which provides a way of thinking for the development of subsequent rehabilitation robots. At the same time, we hope to further optimize and upgrade the developed hand rehabilitation robot in the future. It can be applied to the clinical or household use of hand function rehabilitation.