Posts Tagged active participation

[ARTICLE] Development, Dynamic Modeling, and Multi-Modal Control of a Therapeutic Exoskeleton for Upper Limb Rehabilitation Training – Full Text


Robot-assisted training is a promising technology in clinical rehabilitation providing effective treatment to the patients with motor disability. In this paper, a multi-modal control strategy for a therapeutic upper limb exoskeleton is proposed to assist the disabled persons perform patient-passive training and patient-cooperative training. A comprehensive overview of the exoskeleton with seven actuated degrees of freedom is introduced. The dynamic modeling and parameters identification strategies of the human-robot interaction system are analyzed. Moreover, an adaptive sliding mode controller with disturbance observer (ASMCDO) is developed to ensure the position control accuracy in patient-passive training. A cascade-proportional-integral-derivative (CPID)-based impedance controller with graphical game-like interface is designed to improve interaction compliance and motivate the active participation of patients in patient-cooperative training. Three typical experiments are conducted to verify the feasibility of the proposed control strategy, including the trajectory tracking experiments, the trajectory tracking experiments with impedance adjustment, and the intention-based training experiments. The experimental results suggest that the tracking error of ASMCDO controller is smaller than that of terminal sliding mode controller. By optimally changing the impedance parameters of CPID-based impedance controller, the training intensity can be adjusted to meet the requirement of different patients.

1. Introduction

Over the past decade, the increasing stroke patient population has brought great economic and medical pressures to the whole society. Surviving stroke patients usually have a lower quality of life dues to physical disability and cognitive impairment. Studies on clinical stroke treatment indicate that appropriate rehabilitation training has positive therapeutic effects for avoiding muscle atrophy and recovering musculoskeletal motor functions. However, the conventional one-on-one manual-assisted movement training conducted by physiotherapists suffers from many inherent limitations, such as high labor intensity, high cost, long time consumption, lack of repeatability, low participation levels of patient, and high dependence on personnel with specialized skills [1,2]. In recent years, robot-assisted rehabilitation therapies have gained growing interest from academic researchers and the healthcare industry around the world due to their unique advantages and promising application perspectives. Compared with the traditional manual rehabilitation treatment, the combination of robotic technologies and clinical experience can significantly improve the performance and quality of training. Robot-assisted therapy is capable of delivering high-intensity, long-endurance, goal-directed, and low-cost rehabilitation treatment. Moreover, the functional motivations of patient can be activated to enhance active participation and recover cognitive functions. The physical parameters and therapy data can be recorded and analyzed via sensing system, and that can provide objective basis to optimize training strategy and accelerate recovery process [3,4].
Many therapeutic robot system have been developed to assist stroke patients with motor dysfunctions perform the desired rehabilitation training. The existing rehabilitation robotic devices can be categorized into two types, i.e., end-effector-based robots and exoskeleton-based robots. End-effector-based robot has only a connection between its distal end and the impaired extremity of patient. However, the movement of end-effector cannot uniquely identify the configuration of human limb due to the kinematic redundancy. Miller et al. developed a lightweight and potable end-effector-based therapeutic robot, which is integrated with a wrist and finger force sensor module named WFES, for the upper limb rehabilitation training of hemiplegic stroke patients [5]. Pedro et al. developed a parallel kinematic mechanism (PKM) with two translational and two rotational degrees of freedom (DOFs) for knee diagnosis and rehabilitation tasks [6]. Kang et al. proposed a modular and reconfigurable wrist robot called CR2-Haptic for post-stroke subjects to train forearm and wrist movements [7]. Besides, many other end-effector-based therapeutic robot have been investigated and can be referred to [8,9,10,11,12,13]. Comparatively, the exoskeleton-based rehabilitation robots are developed with more complex structures imitating the anatomical human skeleton and guaranteeing the alignment between the joints axis of robot and impaired limb. ChARMin is a powered exoskeleton integrated with audiovisual game-like interface. It can provide intensive pediatric arm rehabilitation training for the children and adolescents with affected motor functions [14]. Simon et al. proposed a spherical shoulder exoskeleton with a double parallelogram linkage to eliminate singularities and achieve good manipulability properties [15]. Crea et al. developed a semi-autonomous whole-arm exoskeleton for the stroke patients performing activities of daily living (ADL) utilizing hybrid electroencephalography and electrooculography feedback signals [16]. Many other representative exoskeleton-based therapeutic robot have also been designed, such as CAREX-7 [17], RUPERT [18], ULEL [19], ArmeoPower [20], Indego [21], and ETS-MARSE [22].
The effectiveness of robot-assisted rehabilitation training depends on the control strategies applied in the therapeutic robot system. Currently, many kinds of control strategies have been developed according to the requirements of patients with various impairment severities in different therapy periods. The existing control schemes can be basically divided into two categories based on the interaction between therapeutic robots and patients, i.e., patient-passive training control and patient-cooperative training control. During the acute period of hemiplegia, the impaired extremity is fully paralyzed without any muscle contraction. The patient-passive training can imitate the manual therapeutic actions of a physiotherapist. It is especially well suited for the patients with severe paralysis to passively execute repetitive reaching missions along predefined training trajectories. However, it is a challenge to guarantee the position control accuracy during rehabilitation training due to the highly nonlinear properties and unexpected uncertainties of human-robot interaction. Different kinds of control algorithms have been developed to improve control performance of patient-passive training, including neural proportional-integral-derivative (PID) control [23], neural proportional-integral (PI) control [24], adaptive nonsingular terminal sliding mode control (SMC) [25], disturbance observer-based fuzzy control [26], neural-fuzzy adaptive control [27], adaptive backlash compensation control [28], and so on. Comparatively, the patient-cooperation training is applicable for the patients at the comparative recovery period, who have regained parts of motor functions. Clinical studies show that integrating the voluntary efforts of patients into rehabilitation training benefits to accelerate recovery progress and promote psychological confidence. Thus, patient-cooperation training should be able to regulate the human-robot interaction in accordance with the motion intentions and hemiplegia degrees of patients. Many patient-cooperation control strategies have been proposed, such as minimal assist-as-needed controller [29], myoelectric pattern recognition controller [30], electromyography (EMG)-based model predictive controllers [31], subject-adaptive controller [32], and fuzzy adaptive admittance controller [33].
Taking the above into consideration, the contribution of this paper is to develop an upper limb exoskeleton to assist the patient with motor disabilities perform multi-modal rehabilitation training. Firstly, the overall mechanical structure and the MATLAB/xPC-based real-time control system of the proposed therapeutic robot are introduced. Secondly, the dynamic modeling of the human-robot system is researched, and the dynamics parameters are obtained via virtual prototype and calibration experiments. After that, a multi-modal control strategy integrated with an adaptive sliding mode controller and a cascade-proportional-integral-derivative (CPID)-based impedance controller is proposed. The controller is combined with an audiovisual therapy interface and is able to realize patient-passive and patient-cooperation training based on the motor ability of patient. Finally, the effectiveness and feasibility of the developed rehabilitation exoskeleton system and control scheme are verified through three experiments conducted by several volunteers.[…]

Continue —> Sensors | Free Full-Text | Development, Dynamic Modeling, and Multi-Modal Control of a Therapeutic Exoskeleton for Upper Limb Rehabilitation Training | HTML

Sensors 18 03611 g001

Figure 1. Architecture and major components of the upper extremity rehabilitation exoskeleton. (a) Virtual prototype model. (b) Real-life picture of exoskeleton.

, , , , , , , , , ,

Leave a comment

Haptic Control of a Rehabilitation Robot – Full Text PDF

1 Introduction

Robotic support has gained more and more interest in rehabilitation of human haptic behavior, e.g. after stroke. First types of rehabilitation robots were intended to replace repetitive movements performed by a physiotherapist by guiding the patient along a physiological reference trajectory. The robot has the advantage of an accurate and repetitive movement while being resistant to any type of fatigue.

New understanding of motor learning shows that active participation of the patient is an essential element of rehabilitation success. A rehabilitation robot should therefore be just as cooperative as the physiotherapist and enhance the patient’s activity. That means that they should only support the patient if needed. It has also been shown that perturbations such as increasing the error in the patient’s movement can progress the rehabilitation procedure more quickly than only “guiding” the patient to perform the correct movement. This form of therapy has some limitations however, if the patient is not able to apply the necessary forces for the movement. In this case the robot should give appropriate support, for example by providing partial weight support of the patient’s arm if the patient is not able to support their own weight. This simulated weightlessness is able to compensate for muscle disabilities and increase the range of motion during training sessions.

Furthermore, a rehabilitation robot can support the patient during specific tasks by recognizing movement deficiencies and disabilities. The robot supports as much as needed and as little as possible. Such a controller has been implemented in the armrobot ARMin (Figure 1, left). While the user is playing a ping-pong game, the robot is able to support the user as much as needed. In human gait rehabilitation, controller design is more restricted for the sake of security. In the Lokomat (Figure 1, right), path controlling is employed to ensure safe and still self-motivated walking. The path controlling method provides a tunnel for joint angles within which the patient can move. As soon as the patient exceeds the pre-set path trajectory limits, the robot pushes the patient back into the right direction. Figure 2 illustrates and explains the concept of path controlling. Another concept is employed in virtual model control (VMC) which aims at maximum patient activity and only supports selectively chosen characteristics such as length or height of the patient’s stride.

All of those control strategies require the robot to assist-as-needed. The assistance can be interpreted as a virtual helping hand. These virtually created worlds are able to display different forms, from free user-performed movements (no help) to resistance against “wrong” user movements (support), or even guiding the patient through their movement completely. In case of the patient being able to self-perform movements correctly, ideally, the robot should not be felt. This behavior is called transparency.

In addition to movement support, a rehabilitation robot is able to display a virtual world which the user can interact with. This is used for simulating activities if daily living (ADL) such as cooking. The representation of a virtual environment requires the possibility of displaying different virtual objects. Especially hard objects are important. Such requirements for the control of a hard environment differ a lot from those for the control of a free, transparent environment. Two different actuator and controller concepts are optimal to be employed to display a hard or soft environment respectively. The two strategies are called impedance and admittance control and will be the central part of this exercise.

Furthermore, we have to make sure that the human-robot-interaction is safe and secure, i.e. the robot should also be able to navigate a totally passive patient. Therefore, the actuators must fulfill some requirements on power and torque. This includes high transmission ratios, which additionally increase the reflected inertia of the drives. High robot inertia lowers the reachable transparency of the robot. Another important point is backdriveability, which makes the robot movable when the robot is not powered at all. This is an important fact e.g. for the case of an emergency stop.

To sum up, the design and choice of the hardware as well as the software implementation should balance each other. The robot has to bring enough forces and moments to support the patient. A strong (therefore heavy) robot arm is well able to display a hard virtual object such as a wall. On the other hand, the robot should be backdriveable and therefore be as lightweight as possible to easily display transparency. Inertia and mass of a strong (heavy) motor in the system make it difficult to display free environment such as air. Besides the choice of the hardware, the choice of the control strategy is an important fact, too. We will focus on two different strategies of how to display a virtual environment and discuss the concepts of impedance and admittance control.

Full Text PDF

, , , , , , , ,

Leave a comment

%d bloggers like this: