Design, manipulability analysis and optimization of an index finger exoskeleton for stroke rehabilitation

Highlights

• A wearable index finger rehabilitation exoskeleton (WIFRE) is proposed.

• Effects of dimensional parameters of the WIFRE on the manipulability are obtained.

• A multi-objective optimization method for the WIFRE is proposed.

Abstract

Rehabilitation medicine studies have indicated that occupational therapy has a positive effect on the recovery of hand function. However, few wearable hand exoskeletons can assist in performing the adduction and abduction exercises of occupational therapy. Therefore, the wearable index finger rehabilitation exoskeleton (WIFRE) is studied in this paper. First, a WIFRE is proposed to realize the index joint’s independent actuation, including adduction and abduction movements. Second, the local/global kinematic and dynamic manipulability measures are proposed and analyzed to evaluate the performance of the WIFRE. The analysis results show that the dimensional parameters of the WIFRE have a significant effect on its global manipulability measures. The global kinematic and dynamic manipulability measures can be improved by 13%–15% compared with the corresponding minimums. Third, a multi-parameter multi-objective optimization method is proposed to simultaneously enhance the three global manipulability measures. Finally, experiments are performed to verify the effectiveness of the index finger exoskeleton and the global manipulability measures.

Introduction

Cerebral stroke has become a common disease that affects tens of millions of people worldwide, especially people over 40 years old [1], [2]. This disease may cause the impairment of stroke patients’ motor functions and bring lots of trouble to their activities of daily living. Conventional rehabilitation training for regaining their motor functions requires therapists’ assistance, which is a labor-intensive and repetitive task for therapists. Therefore, there is a strong need for rehabilitation robots to assist, improve and assess the rehabilitation training process for stroke patients [3], [4], [5]. Many clinical trials have been validated that hand rehabilitation robots are feasible and effective for the rehabilitation therapy of stroke patients [6], [7].

Hand rehabilitation exoskeletons can be divided into soft exoskeletons, soft-rigid exoskeletons, and rigid exoskeletons. Fischer et al. proposed a cable-driven soft glove to assist in finger extension and complete the clinical assessment [8]. Yurkewich et al. proposed a soft tendon-driven hand extension robot orthosis glove to enable stroke survivors to grasp and stabilize objects [9]. By taking advantage of the soft exoskeletons, the soft-rigid exoskeletons have been proposed to improve the output forces. For example, Bos et al. developed a novel hydraulic hand exoskeleton with flexure elements to minimize the shear forces and enhance its comfort [10]. Nycz et al. designed a soft hand exoskeleton with artificial tendons to assist and restore hand functions of stroke patients [11]. However, the soft exoskeletons and soft-rigid exoskeletons have an inherent disadvantage: they usually cannot provide enough output forces in rehabilitation exercises. Therefore, this paper focuses on the study of rigid linkage-type exoskeletons.

The design of rigid linkage-type exoskeletons is challenging due to the complexity and the inter-subjective variability of human hand musculoskeletal. Many studies have been conducted to deal with the design problems of rigid linkage-type exoskeletons. Zhang et al. proposed a hand rehabilitation exoskeleton using symmetrical pinions and racks to adapt to fingers of different sizes and avoid secondary injuries [12]. Jo et al. proposed a portable hand exoskeleton with one degree of freedom (DOF) to exercise finger flexion/extension movement [13]. Wang et al. proposed a finger rehabilitation robot using pneumatic muscles and controllable magnetorheological dampers for active and passive rehabilitation training [14]. Chiri et al. developed a novel wearable multi-phalange hand exoskeleton for post-stroke rehabilitation, which can minimize the human exoskeleton axes misalignment [15]. Agarwal et al. proposed and fabricated an index finger exoskeleton with series elastic elements based on a linkage mechanism [16]. Hong et al. proposed an underactuated hand exoskeleton for power assistance of the hand using a spherical four-bar linkage [17]. However, the aforementioned index finger exoskeletons can only complete the flexion/extension (f/e) movement. Since the human index finger’s natural movements, related to f/e and abduction/adduction (a/a), are spatial movements, the exoskeletons that can assist f/e and a/a should be further studied.

To show the significance of the index finger exoskeleton with both f/e and a/a, the hand exoskeleton with only f/e and that with both f/e and a/a are compared here. First, the rehabilitation medicine studies have shown that breaking down a complex movement into simpler, anatomically isolated movements could produce a better rehabilitation effect [18]. That is, the index finger exoskeleton with both f/e and a/a could produce a better rehabilitation effect compared with the hand exoskeleton with only f/e. Second, the studies have shown that recovery of the normal kinematic coordination is essential and valid for sensorimotor training [19]. Since the exoskeleton with both f/e and a/a can complete the coordinated movement of f/e and a/a, this exoskeleton may have a better rehabilitation effect than that with only f/e [20]. Third, rehabilitation medicine studies have indicated that occupational therapy has a positive effect on recovery of hand function [21]. According to occupational therapy, at least for the last step of rehabilitation exercises, stroke patients should perform some delicate movements, such as turning knobs, handling chopsticks, grasp spherical objects, which need the coordination of f/e and a/a [22], [23]. Therefore, the finger exoskeletons should assist the finger f/e and the a/a movements independently.

Many researchers have made efforts to assist a/a and f/e simultaneously. Cempini et al. proposed a general method for design and analysis of two-DOF self-aligning mechanisms based on a planar decomposition approach and designed a cable-driven index finger exoskeleton [24], [25]. However, this exoskeleton can only actuate the f/e movement of the metacarpophalangeal (MCP) joint actively since there is no actuator for a/a movement. Ueki et al. proposed a multi-DOF rehabilitation robot able to assist f/e and a/a of the index finger [26]. However, this hand rehabilitation robot is fixed on a heavy arm holding part; thus, it is not convenient to assist the activities of daily living. Agarwal et al. proposed a four-DOF thumb exoskeleton to achieve a large range of motion and allow for bidirectional torque control of each thumb joint individually [27]. However, this exoskeleton requires that the axis of the exoskeleton a/a movement is aligned with the axis of the thumb a/a movement, which is difficult to achieve for a wearable device. According to the authors’ knowledge, the index finger exoskeletons that can assist the f/e and the a/a movements independently remain few. Thus, this paper aims to propose such an index finger exoskeleton to satisfy this requirement for hand rehabilitation.

Robot’s performance improvement is critical for the finger exoskeleton. In this paper, the robot’s optimal design is studied based on manipulability, which is an important performance index for evaluating the positioning and orienting capabilities of the robot’s end-effectors. Yoshikawa et al. first proposed the concept of the manipulability and manipulability ellipsoid in [28]. Chiu et al. proposed the task compatibility to evaluate the robotic mechanism’s manipulability using a task space description [29]. Chiacchio et al. proposed the force manipulability ellipsoid of redundant robots to evaluate the force capability under a static assumption [30]. Chiacchio et al. proposed a dynamic manipulability based on the dynamic model to evaluate the end-effectors’ accelerations [31]. Azad et al. studied the weighting matrix of the dynamic manipulability to measure the ability to accelerate the robot end-effector [32]. Zhou et al. analyzed the dynamic manipulability of the multi-arm space robot, and obtained the effects of link lengths and joint variables on the dynamic manipulability [33]. Lachner et al. analyzed the effects of the coordinates on the dynamic manipulability, and proposed a manipulability measure as a property of the physical object [34]. The kinematic and dynamic manipulability measures have been researched to analyze or optimize different robots in recent decades [35], [36], [37], [38]. However, the kinematic and dynamic manipulability measures have not been explored to analyze or optimize a wearable hand exoskeleton mechanism that forms a closed-chain together with the human hand.

In this paper, a wearable index finger rehabilitation exoskeleton (WIFRE) is proposed to assist the f/e and a/a exercises independently. Its manipulability is analyzed, and its optimal method is proposed. The contributions of this paper can be summarized as follows.

(1) A novel WIFRE is proposed to independently assist in performing the f/e and a/a exercises of the finger and eliminate the shear force on the finger phalange.

(2) Local/global manipulability measures of the WIFRE are analyzed, and effects of the WIFRE dimensional parameters on the manipulability measures are obtained.

(3) An optimization method of the WIFRE is proposed to obtain the optimal dimensional parameters through solving a multi-parameter multi-objective optimization problem.

The paper is organized as follows. In Section 2, a WIFRE is designed. In Section 3, the kinematics of the WIFRE is established, and the kinematic manipulability is analyzed. In Section 4, the dynamics are established, and the dynamic manipulability is analyzed. In Section 5, the optimization method for the WIFRE design is proposed. Section 6, experiments and discussion are carried out. Finally, Section 7 gives conclusions and outlook.