Jiting Li

Design of an exoskeleton for index finger rehabilitation

This paper presents a new exoskeleton with 4 degrees of freedom (DOF) for index finger rehabilitation. The device can generate bi-directional movement for all joints of the finger through cable transmission, which is required for passive and active trainings. With two prismatic kinematic joints in the design, it can accommodate to some extent variety of hand sizes. The kinematic relation between the device joint angles and the corresponding finger joint angles is simple which greatly simplifies the high level motion control. As the motor capability of patients may be different and the range of motion of the finger may change along with the rehabilitation progress, it is important to take the changes into consideration. And the preliminary experiment has shown that the proposed device is capable of accommodating to these varieties.

iHandRehab: An interactive hand exoskeleton for active and passive rehabilitation

This paper presents an interactive exoskeleton device for hand rehabilitation, iHandRehab, which aims to satisfy the essential requirements for both active and passive rehabilitation motions. iHandRehab is comprised of exoskeletons for the thumb and index finger. These exoskeletons are driven by distant actuation modules through a cable/sheath transmission mechanism. The exoskeleton for each finger has 4 degrees of freedom (DOF), providing independent control for all finger joints. The joint motion is accomplished by a parallelogram mechanism so that the joints of the device and their corresponding finger joints have the same angular displacement when they rotate. Thanks to this design, the joint angles can be measured by sensors real time and high level motion control is therefore made very simple without the need of complicated kinematics. The paper also discusses important issues when the device is used by different patients, including its adjustable joint range of motion (ROM) and adjustable range of phalanx length (ROPL). Experimentally collected data show that the achieved ROM is close to that of a healthy hand and the ROPL covers the size of a typical hand, satisfying the size need of regular hand rehabilitation. In order to evaluate the performance when it works as a haptic device in active mode, the equivalent moment of inertia (MOI) of the device is calculated. The results prove that the device has low inertia which is critical in order to obtain good backdrivability. Experimental analysis shows that the influence of friction accounts for a large portion of the driving torque and warrants future investigation.

Development of a Hand Exoskeleton System for Index Finger Rehabilitation

In order to overcome the drawbacks of traditional rehabilitation method, the robot-aided rehabilitation has been widely investigated for the recent years. And the hand rehabilitation robot, as one of the hot research fields, remains many challenging issues to be investigated. This paper presents a new hand exoskeleton system with some novel characteristics. Firstly, both active and passive rehabilitative motions are realized. Secondly, the device is elaborately designed and brings advantages in many aspects. For example, joint motion is accomplished by a parallelogram mechanism and high level motion control is therefore made very simple without the need of complicated kinematics. The adjustable joint limit design ensures that the actual joint angles don’t exceed the joint range of motion (ROM) and thus the patient safety is guaranteed. This design can fit to the different patients with different joint ROM as well as to the dynamically changing ROM for individual patient. The device can also accommodate to some extent variety of hand sizes. Thirdly, the proposed control strategy simultaneously realizes the position control and force control with the motor driver which only works in force control mode. Meanwhile, the system resistance compensation is preliminary realized and the resisting force is effectively reduced. Some experiments were conducted to verify the proposed system. Experimentally collected data show that the achieved ROM is close to that of a healthy hand and the range of phalange length (ROPL) covers the size of a typical hand, satisfying the size need of regular hand rehabilitation. In order to evaluate the performance when it works as a haptic device in active mode, the equivalent moment of inertia (MOI) of the device was calculated. The results prove that the device has low inertia which is critical in order to obtain good backdrivability. The experiments also show that in the active mode the virtual interactive force is successfully feedback to the finger and the resistance is reduced by one-third; for the passive control mode, the desired trajectory is realized satisfactorily.

Multiple rehabilitation motion control for hand with an exoskeleton

This paper investigates the control algorithm of an exoskeleton for hand rehabilitation, which can realize the active, passive, and assisted rehabilitation motion. The active mode is accomplished with the force control algorithm during which the resistance is compensated in free space and the virtual interactive force is rendered to the finger in constraint space. The passive mode is realized by the position controller given the desired motion trajectory. The assisted mode is implemented in the customized positions by switching between the active and passive modes according to the predefined action procedure. The experiments are conducted to verify the proposed method, and the results show that the different rehabilitation motion are successfully accomplished and the maximum joint position error is less than 1.2 degree, which satisfies the requirement in hand rehabilitation application. The results demonstrate the validity of the proposed method.

Active and Passive Control of an Exoskeleton with Cable Transmission for Hand Rehabilitation

Abstract —This paper investigates the control algorithm of an exoskeleton for hand rehabilitation, which accomplishes both active and passive control mode. A double closed loop control structure is developed, which consists of position control loop and compensation control loop. The position controller is based on impedance control. The compensation controller is used for compensating the position error caused by deflection of the cable and sheath in the mechanical transmission. To realize the compensation, the spring model is used to represent the elasticity of the cable and sheath. With the proposed method, the maximum joint position error is about 1.5 degree, which satisfies the requirement in hand rehabilitation application. The experimental result demonstrates the validity of the propose method. Keywords- rehabilitative training; active control mode; passive control modes ； compensation controller I. I NTRODUCTION As we know, the normal motor capability of hand is crucial and important for human-being’s daily life. Hands, however, are apt to be injured in accident. And the rehabilitation is essential for the patients to recover after hand operation. Additionally, diseases, stoke for instance, can also result in the loss of hand function. In order to regain the motor capability, the hand rehabilitation is a fundamental therapeutic approach. The traditional rehabilitation approach is costly for patients and laborious for therapists. Recent research showed that hand rehabilitative training using mechatronic devices and virtual reality is possible and effective [1] and is attracting more research interests [2-8]. Dependent on different design and different application, some control algorithms are investigated [6-11]. Despite of the researches, there are still some problems to be investigated. The rehabilitation usually includes four modes, i.e. passive, active, assisted and resisted rehabilitation. So far, there is no solution which has covered all of the four control modes. Aimed at the hand rehabilitation, our research group developed a wearable exoskeleton for index finger rehabilitation [12]. In this paper, the control strategy of this exoskeleton is investigated to provide both active and passive control modes. A compensation control method is presented to reduce the position error due to deflection of the cable and sheath in the mechanical transmission. To do this, a double closed loop structure is developed, which is used for realizing the position control and error compensation, respectively. With the proposed method, the maximum joint position error is less than 1.5 degree, which satisfies the requirement in hand rehabilitation application. The remainder of the paper is organized as follows. Section 2 introduces the mechanical structure of the exoskeleton. Section 3 describes the control algorithm of active and passive control modes. Section 4 depicts experiment and the results. Section 5 gives conclusion and future work. II. M

A resistance compensation control algorithm for a cable-driven hand exoskeleton for motor function rehabilitation

The resistance compensation, especially the friction compensation in the Bowden cable transmission is a difficult issue to be handled. Aimed to the resistance reduction requirement in the active rehabilitative motion, a resistance compensation control method is proposed. Based on the simplified transmission model, the resistance, including the cable friction as well as the mechanical moment of inertial, is formulated. To realize the compensation, force sensors are used to measure the force exerted by the human fingertip. With the proposed algorithm, the maximum finger-exerted force is reduced to less than one third of before. The experimental result demonstrates the validity of the proposed method.

Active and passive control algorithm for an exoskeleton with bowden cable transmission for hand rehabilitation

This paper investigates the control algorithm of an exoskeleton for hand rehabilitation, which accomplishes both active and passive rehabilitation training. In the passive mode control the PID control algorithm is executed in the velocity mode of the driver. In the active mode control, control architecture is proposed to deal with in both free space and constraint space. A resistance compensation control method is proposed to reduce the resistance in free space which is caused by the friction of the Bowden cable as well as the moment of inertial. To realize the compensation, force sensors are used to measure the force exerted by the human fingertip. A commercial driver, which could switch between the two control modes by a programmable digital switch rather than changing the physical connection manually, guarantees the realization of the required functions. The experiments are conducted to verify the proposed method, and the results show that in the active control mode, the maximum finger-exerted force with compensation is about two fifths of the force without compensation which means the resistance is greatly reduced. And in the passive mode, the maximum joint position error is about 1.2 degree, which satisfies the requirement in hand rehabilitation application. The experimental results demonstrate the validity of the proposed method.

Haptic Rendering of Virtual Hand with Force Smoothing

In this paper, we present methods to generate a stable and realistic force rendering between virtual hand and object for a hand rehabilitation system. There are multi-contact regions between hand and object. For each contact region, the virtual contact force is modeled upon the spring-mass model. As direct rendering with spring-mass model in a large stiffness will induce instability of the haptic device, we apply a force-smoothing method by limiting the maximum force variation to guarantee the stability in the haptic rendering. The system experiment results demonstrate that the proposed method can provide satisfactory stable haptic display. And the maximum virtual stiffness that the system can simulate increases from 0.7 N/mm to 1.6N/mm by using force smoothing process.

Real-time deformation simulation of hand-object interaction

This paper investigates the deformation simulation of hand-object interaction for the virtual rehabilitation system of hand. The deformation along the contact normal on hand and object is investigated. Following the anatomy of human hand, the virtual hand organizational structure is simplified as a 2-layered representation, of which the inner layer models the rigid skeleton, and the outer layer represents the deformable soft tissues. The object could be either rigid or elastic. This model is aimed not only to provide the visual realism, but also to enhance the force fidelity for the future work. To realize rapid computation, the deformation of the hand and the object is determined with geometry-based method. To handle the multi-point hand-object interaction, a weighted function is proposed to take the effect of the multiple contact regions on the object deformation into consider. The experimental results show that the deformation is achieved but the further improvement is still needed. The computation time is less than 1ms, which highly satisfies the real-time requirement in our virtual rehabilitation system with force rendering.