Technology and Health Care 19 (2011) 483–495 DOI 10.3233/THC-2011-0646 IOS Press

483

Rehabilitation Engineering

Rehabilitation robotics
Marko Munih and Tadej Bajd∗
Faculty of Electrical Engineering, University of Ljubljana, Slovenia
Abstract. The paper presents the background, main achievements and components of rehabilitation robotics in a simple way, using non-technical terms. The introductory part looks at the development of robotic approaches in the rehabilitation of neurological patients and outlines the principles of robotic device interactions with patients. There follows a section on virtual reality in rehabilitation. Hapticity and interaction between robot and human are presented in order to understand the added value of robotics that cannot be exploited in other devices. The importance of passive exercise and active tasks is then discussed using the results of various clinical trials, followed by the place of upper and lower extremity robotic devices in rehabilitation practice. The closing section refers to the general importance of measurements in this area and stresses quantitative measurements as one of the advantages in using robotic devices. Keywords: Robot, haptic interface, virtual reality, measurement

1. Introduction The application of robotic approaches in neurological patient rehabilitation was introduced almost two decades ago [1]. Even though the number of robotic rehabilitation systems is large, the number of clinical trials remains quite limited. In fact, it is not yet clear what characteristics should be incorporated in a therapeutic robotic assistant platform. Conventional therapeutic techniques and robot assisted techniques must not be perceived as two opposing modalities, but rather as two complementary approaches. Two very positive aspects of robotic therapy are high repeatability and automatic measurement during exercise. In contrast, the activities of a therapist unavoidably include many subjective elements, but an experienced therapist has an in-depth understanding of the individual patient which no high-tech device can ever possess. In future, robotic therapy will complement existing clinical practice: by reducing a therapist’s workload, providing less costly and more extensive therapeutic programmes; by using quantitative measures of an intervention or injury and last, but not least, by new insights into the treatment process. One of the natural common points of conventional and robotic therapy is related to haptic interaction between a patient and therapist. This approach gains specific importance for instance in the Bobath concept, which is based on an holistic approach to a patient, together with the International Classification of Functioning, Disability and Health (ICF) of the World Health Organization. At present, the emphasis is directed to problem solving, on predefining some intermediate goals and then adjusting theraphy in gradual steps to finally learn complete movement, for example, in spasticity. This approach highlights
∗

Corresponding author. E-mail: bajd@robo.fe.uni-lj.si.

0928-7329/11/$27.50  2011 – IOS Press and the authors. All rights reserved

484

M. Munih and T. Bajd / Rehabilitation robotics

and relies on the plasticity of the nervous system. Solving problems includes activities focused at a specific task, motivation, planning, interaction with the environment, selective placement of attention by stimuli and biomechanical aspects for effective and functional direction of movement. Thus, haptic interaction between a patient and a therapist must not be invasive and unilateral. On the contrary, it must evoke all the capabilities of a person leading to functional movement by minimum intervention. To enable the best possible imitation of natural human interaction, while using a machine as a rehabilitator assistant, a robotic therapist must be highly consistent with human interaction. It is no coincidence that the original ideas which brought these systems to life did not arise in the rehabilitation field, but rather in neurophysiology and haptic perception, including sensimotor learning [7–9]. Robotic therapist devices differ in a number of ways from industrial robots. These are traditionally moved between two known points along a defined trajectory. Maintaining a known and highly accurate position is essential. In contrast, the robotic therapist must be programmable and adjustable. This can be achieved by impedance control algorithms or by passive manual control as in backdrivable robot manipulators including MIT-Manus and Braccio di Ferro. Other devices designed for robotic therapy might use industrial manipulator technology and have high manipulator arm mechanical impedance due to high gearing ratio and admittance control scheme. Robotic devices may be used in positional trajectory mode for enforcing passive movements, for simple positional control of extremities, which, does not involve active participation of a patient on either neuromuscular or sensory levels. Such passive exercise, quite the opposite of the Bobath concept, using a robot contributes to rehabilitation, at least in specific clinical conditions [10–12]. Other studies [13–15] indicate better results with the application of techniques that consider the adaptive nature of the nervous system by solving problems with a suitable adaptability level. These techniques include active assisted exercises in which the robot moves the extremity along a predetermined trajectory. In the active constrained exercises, the robot provides higher opposing forces. The movement may be limited inside some 3D virtual space, with opposing forces applied when the subject tries to move outside this region. Similarly, one could imagine a ball object that limits movement toward the center (inside), with completely empty space all around. Furthermore, in active resistive exercises, the robot opposes the intended movement. In adaptive exercises, the robot is providing a previously unknown dynamic environment to which the subjects have to react and adapt. Passive exercises need no input from the patient, while active constrained exercises require at least residual movement cababilities or subconscious sensory-motor co-operation. Active resistive and adaptive exercises require cooperation with sufficient volitional motor activity. Active resistive and adaptive exercises are therefore not a suitable choice for persons with larger movement deficits, since they usually cannot use them independently. Still, such persons may exploit robot-therapeutic exercises in which minimum co-operation on their side will help them in using and strengthening their remaining abilities. 2. Virtual reality in rehabilitation Most of us are familiar with virtual reality (VR) technology from entertainment (e.g. games) and military simulations. Lately, its use has expanded to other areas, for instance computer-aided design (CAD), architecture, general virtual presentation of data and to medical applications. Medical applications of virtual reality include training in medicine and surgery, modelling of hard and soft tissue, image displays, remote surgery, ergonomy and rehabilitation. In rehabilitation, virtual reality is used for training and measuring the motor ability (e.g. in hemiplegia, paraplegia, Parkinson’s disease).

M. Munih and T. Bajd / Rehabilitation robotics

485

Virtual reality is usually understood as a three-dimensional computer model which primarily defines the geometrical model (kinematic or dynamic model) of different virtual objects and their environment. It is possible to define not only the image of objects, but also their inherent physical characteristics. Such a virtual world can change in time – not only the position and orientation of particular items, but also dynamic characteristics of the surroundings, such as friction and gravity. Realistic visualisation and deep presence of person in VR are mostly dependent on high-quality graphics which take the user into the virtual world. The impression of virtual reality depends on the realistic appearance of the scene and activity. Simpler non-immersive visualisations include a 2D computer screen, a projection screen and 3D projection techniques in the environment. The immersive method is more realistic, enabling the user to percieve a full 3D visual field by use of special glasses. The dynamic characteristics of objects, including mass, moment of inertia, forces, torques, and torques and forces resulting from the interaction of objects and, for instance, compliance (stiffness) of objects, roughness and smoothness of objects cannot be visually detected or presented. These require the sense of touch. 3. Hapticity The verb ‘απτ ω ’ – hapto is originating from Greek for reaching, holding and touching. If a person exerts a force F onto a mass m, which is integrated in the environment via the damper h and the spring k, the movement x is determined by the differential equation:
F = m d2 d + k + h 2 dt dt x = mx + hx + kx ¨ ˙

(1)

If the coefficients m, h and k are constant, the force depends on the position (distance to origin), velocity and acceleration. The first force term in the equation depends on mass and acceleration m x, ¨ with m representing the mass of e.g. a cube on icy surface. Significant force will be needed only during acceleration and deceleration with high m, while constant velocity movement requires no force. Only a small force is needed for the acceleration and deceleration of a light styrofoam cube with a small m. The force can, as a second example, depend on the term h x, where h stands for damping of e.g. of an oar in ˙ water. If the velocity x is low, only a small force is needed, whereas at higher velocities, a considerably ˙ greater force is required. Another medium, like oil or a spoon in honey or air, represents a different ˙ damping coeficient value h and thus a different force at the same velocity x. The damping force is small when an oar moves in air, but if the velocity is high, in the case of an airplane propeller, the pull/push force becomes significant. The impact of stiffness k is presented by means of a spring. The greater the deviation x, the greater the force. Also, a higher spring stiffness coeficient k value, with x unchanged, requires greater force. A general body in a virtual environment, into which a person exerts a force, is not a point mass, pure damper or spring but rather a combination of all three terms in the above equation that becomes more complex. This is typical for haptic touch in a virtual environment, where none of the three coeficients of the equation is constant. The values of m, h and k change locally in the virtual environment. Actually, one of the three parameters, e.g. k, has six values at the same point (or position) in space: kx , ky and kz along individual axes of space, and kRx , kRy and kRz around individual axes of rotation – thus it has six degrees of freedom (DOF) (Fig. 1). An empty space has nil or small values of all six k. A flat wall in a virtual environment is represented as a high k horizontally. A wall can also be slippery as

486

M. Munih and T. Bajd / Rehabilitation robotics

Fig. 1. Second-order mechanical system with six DOF, three translations and three rotations and m, h and k.

ice or non-slippery if coated with rubber, which is set with the other two (transversal) k parameters of the wall. The remaining three k elements stand for the three rotational stiffness coeficients. Similar six-dimensional understanding applies to the other two parameters – h and m. In rehabilitation haptic rendering is often used to denote a virtual tunnel that connects two extreme points of movement, the starting and the final one. The trajectory between them can run over a straight line or a curve having a different shape. All elements m, h and k along the direction of movement equal zero, whereas, perpendicular to the direction of movement, the stiffness coefficient k increases by selected functions. This is reflected in the virtual pipe or tunnel that forces the user to the central curve line, where this force component does not exist. In a series of attempts, should the user appear at exactly the same coordinate of space, carrying the same coefficients, the current values of x and x (current acceleration and velocity) are very likely to ¨ ˙ differ, meaning that the force F felt by the user in a virtual environment with a haptic interface is also different. A haptic interface is, therefore, a robot capable of functioning in line with the above equation, contrary to the classic understanding of robot functioning. These are usually position-controlled, as referred to under passive exercise. Haptics in a virtual environment is achieved in real time by setting the interaction force in the point between the user and interface, considering all the numerous parameters mentioned above. It is understandable why this technology has become accessible only in the last two decades. The connection of quantities F and x represents an energy contact, in reality the transfer of power between man and environment F x and thus controlled transfer of energy via man-machine interaction. In ˙ addition to the vision sensory pathway, the user participating in haptic interaction in a virtual environment also engages and uses touch, position, force and texture sensation in their body as well as the mechanisms of visual and tactile recognition, along with reflex and volitional control mechanisms, including the entire motor chain.

M. Munih and T. Bajd / Rehabilitation robotics

487

Further to vision and haptic modalities, further information can be supplied to the user via sound. Particular cases would be sound produced during movement along rough or smooth surfaces, along a ladder, or when typing on virtual keyboards. 4. Passive exercise CPM (Continuous Passive Motion) devices have been used in the postoperative rehabilitation of joints since approximately 1960. It was later established that joint immobilisation in a test group of rabbits led to great problems in restored mobility [16]. After five to six weeks of immobilisation, most joints developed moderate to extreme changes, including those in joint cartilage and bone mass, with changes already apparent in the second week. In clinical practice, this presented as joint contractures and reduced range of motion (ROM). To avoid this, they tried to maintain a good ROM by simple CPM devices which moved the limb over an arc corresponding to joint movement. Such devices consist of a simple motor with a mechanically or electronically adjustable range and velocity of movement. Most modern CPM devices adopt ideal joints with a fixed point of rotation, and can provide movement in one vertical plane (2D). Usually they cannot be reprogrammed and are controlled using the open loop principle, which means that their movement angles and forces are not measured immediately but are corrected and reset a few 100-times in a second. Such CPM devices are very efficient in preserving range of movement, they reduce stiffness in joints, decrease the need for drug administration and shorten the length of hospitalization. Comparison of CPM and physiotherapy reveals no major differences as regards the above parameters, however in the majority of cases, CPM results in weaker muscles, delays in activation of extensors and stiffness of flexors [17]. It would be difficult to expect anything else, since the CPM devices only move a person’s extremities without activating the muscles. 5. Active exercises Several research groups developed robotic devices for rehabilitation of the upper and lower extremities. These devices, designed for exercise, use various methods of operation. Active constrained movements are made possible by most of the mentioned devices, while movements with active resistance are provided by MIT-Manus, Bi-Manu-Track and MIME [18–20]. Adaptive exercises are possible using Bi-ManuTrack and MIME [19,20], in which case the healthy arm leads and the injured arm imitates movement, both moving simultaneously. The literature often contains descriptions of technical approaches and solutions. However, there is a limited number of clinical trials with clear goals and methodology that would systematically present the influence of robotic devices and existing modes of operation on the rehabilitation process. Among 17 known clinical trials, for a variety of reasons, only a handful can be used for direct comparison. None of them systematically investigates the influence of the method of applied exercise. Most trials have so far employed all three modes of operation, including: passive, active constrained and active resistive, regardless of the probability that one of these methods might be more successful than others. Only Fasoli et al. [27] and Stein et al. [28] tried to specifically evaluate particular exercise versions. The findings of the Fasoli trial disclose that active resistive exercises are more useful than active constrained exercises as regards the upper extremities. Still, the repeated trial covering the same group did not reveal any differences between the exercises involving active constrained movements and those involving active

488

M. Munih and T. Bajd / Rehabilitation robotics

Fig. 2. Trajectory in active constrained exercise.

Fig. 3. Example of environment in active constrained exercise, GENTLE/S project.

resistive movements, perhaps because of the manner of data calculation. In the first trial Fasoli et al. [27] included in the group of an active constrained movement those who did not have sufficient motor function for the active resistive group. This was not the case in the trial conducted by Stein et al. [28]. Thus, the results of active resistive exercises were overrated in advance. These findings point to the fact that some robot-assisted therapies are less suitable for specific groups of patients. Therefore, the precise contribution of individual exercise methods to the rehabilitation of upper extremities in a hemiplegic person remains unexplained. Other methods of exercise are perhaps equally or even more important for robotic rehabilitation, but they have not been investigated sufficiently, or at all. One such method is gravity compensation of the upper extremity. Most robotic devices provide some sort of support for the arm. The evolution of gravity compensation devices has been going on for decades – for example Sanches [29]. Beer et al. specifically investigated the implementation and reach of gravity compensation for the upper extremity [30–32]. Their preliminary research showed that poor coordination of muscle activation led to unexpected torques in the hemiplegic arm joints (e.g. shoulder abduction prevented the ability of elbow extension). Further trials revealed immediate improvement of motor abilities in persons after a stroke, if gravity did not influence the upper extremity. The active abduction of the shoulder was reduced and consequently extension in the elbow increased in static conditions. The latest results point to a similar mechanism

M. Munih and T. Bajd / Rehabilitation robotics

489

Fig. 4. Example of active exercise with active resistance (torus), HIFE project.

Fig. 5. Example of active exercise with active resistance (pipe with a spring), I-Match project.

functioning in dynamic circumstances. Such findings indicate improved motor abilities due to the use of gravity compensation. The second trial involving exercise with a sling revealed no significant difference. Even though the principle of gravity compensation is at present rather unexploited, faster recovery can be expected along with new trials. Recently, three arm gravity compensators were put on the market: Armon1 , Dynamics Arm Support2 (DAS) and Armeo3 . The first two devices provide lower arm support, while the third is an exoskeleton type mechanism allowing free arm movement for people with little muscular arm strength, using passive principles, thus providing for static and not accounting for dynamic compensation. More research in this field can clarify the mechanisms of the compensation impact, as
1 2

www.mginside.info. www.exactdynamics.com. 3 www.hocoma.ch.

490

M. Munih and T. Bajd / Rehabilitation robotics

well as indicate where it would be reasonable to apply these mechanisms. Besides the above approaches, other methods of rehabilitation can additionally promote fast recovery in hemiplegic patients. Interesting assertions were published by Kahn et al. [35], claiming that equally positive changes in motor function were observed in a group undergoing robot-assisted reaching exercise (target is always reached) and in a group undergoing robot-assisted reaching exercise without obligatory reaching of the target. Some other possible techniques deserve mention: functional electrical stimulation, pharmacology, intensive exercise, including many repetitive exercises, loading therapy – automated use of this method – and sensor-motor exercise. A systematic review of the effect of robot-aided therapy on recovery of the hemiplegic arm after stroke, collecting results from eight selected studies is provided by Prange et al. [36]. This indicates that robot-aided therapy of the proximal upper limb can improve short- and long-term motor control of the paretic shoulder and elbow. This statement is supported by quantitative analysis of short-term pooled data in chronic stroke patients and indicates that increased motor recovery of chronic patients is possible after robot-aided therapy. However, no consistent effect on improvement in functional ability has been reported, although the training modalities were not directly designed for this. Restoration of motor control appears greater after robot-aided therapy than conventional tehrapy. It was not possible to establish which aspects of robot-aided therapy (e.g. increased intensity of movements, more effective training modalities) were responsible for the beneficial influence on recovery. Clinical experience seems to show that robot-aided therapy can improve motor control of hemiplegic upper limbs, perhaps even more than conventional therapy, in both sub-acute and chronic stroke patients. 6. Devices for robot-aided therapy A number of research or commercial platforms exist today designed specifically for the tasks of rehabilitation robotics, while in some other platforms the primary design goal has been aimed at something else, but the resulting device also suits the needs of rehabilitation. Regardless of their origin, the devices can be of either the exoskeleton or the end-effector type. An exoskeleton is an external framework, close to, or in contact with, human limbs. Exoskeletons contain rigid and resistant components for each of their segments, being able to providing passive or active translation or rotation in a particular joint. The rotation axes of the robot must correspond to the rotation axes of the human skeleton, and the limb may be connected to the exoskeleton at several points. Powered exoskeletons may be designed, for example, to assist and protect soldiers and construction workers, or to aid the survival of people in other dangerous environments. Wider medical markets exist for providing mobility assistance and rehabilitation for aged and infirm people. A weak point in exoskeleton devices is the limited amount of Z-bandwidth and the transparency of the haptic interface – the parameters that tell how good is the touch (impedance) compared to the ideal model in the virtual environment. A strength is the good fit to each human body segment and the ability to move each limb and interact with supervised known parameters. Perry et al. [39] is using an exoskeleton robot with seven degrees of freedom in the most important joints of human arm. L-Exos is a tendon-driven, wearable haptic interface with 5 DoF, optimized to obtain a solution with reduced mass and high stiffness, by employing special mechanical components and carbon fiber structural parts [41]. Neural control of an upper limb powered exoskeleton system has 8 DoF [44]. ARMin represents an interesting later design that at the moment enables movements in 6 DoF [26].

M. Munih and T. Bajd / Rehabilitation robotics

491

Fig. 6. ARMin robot is one of the recent exoskeleton devices (ETH Z¨ rich). u

In most cases, robotic haptic arms that interact with the environment only via end- effectors are in contact with the human arm via the wrist or with the human leg via the foot only. As a consequence, the trajectory of each body segment under motion is not supervised and determined by a machine, as above, but also depends on the person. This can be good, but may also lead to a less determined environment. Good points in the end-effector approach are better values for Z-bandwidth and transparency of the haptic device. Examples of end-effector upper extremity devices include MIT Manus [18], Assisted Rehabilitation and Measurement (ARM) Guide [19], Mirror Image Motion Enabler (MIME) [20], Bi-Manu-Track [21], GENTLE/S [22], Neurorehabiltation (NeReBot) [23], REHAROB [24], Arm Coordinating Training 3-D (ACT3D ) [25], Braccio di Ferro, [42], NEDO project device [43] and Baltimore Therapeutic Equipment Co. (BTE), produce several training devices. Most devices have been designed for training of the proximal joints (shoulders and elbows) of the hemiplegic arm. Some allow for two-dimensional plane movement only, while in most cases, movement is possible in three dimensions within the limited section of the entire working area of the arm. On the contrary, Bi-Manu-Track includes the forearm and wrist, which is enabled also by the recent version of the MIT-Manus device. Over and over again we encounter new robotic devices and evolutional upgrades of the existing ones (e.g. Furusho et al. [33] and Colombo et al. [34]). There are also devices available for the lower extremities, Gait Trainer GT I [37] and HapticWalker [38] most frequently in combination with treadmills, and also separately for ankles. In the lower extremities, the end-effector devices only guide foot motion, while the multiple degrees of freedom of the rest of the body (e.g. leg, hip) remain completely unrestricted. In the case of a patient with an unstable knee joint, the physiotherapist has to stabilize the knee manually. Studies on the Gait trainer GT I revealed comparable muscle activation patterns as in treadmill walking. Research devices are also available for exercising some other joints of the body, like ankle, wrist, individual fingers [45] or several fingers simultaneously.

492

M. Munih and T. Bajd / Rehabilitation robotics

Fig. 7. GENTLE/S project rehabilitation environment as an example of end effector approach.

7. Measurements in rehabilitation In neurology, standard rating scales are used for measurements, mainly for the purpose of early systemisation. On the other hand, uniform, quantitative, traceable procedures for measurements in rehabilitation have not yet been introduced. One of the models supposedly establishing a logical, understandable and thorough system in rehabilitation was adopted by the World Health Organisation in 1980. Even though the model, known as ICIDH (International Classification of Impairments, Disabilities and Handicaps) is very detailed and subjective (rheumatologic diseases), it provides a solid framework for understanding and treating neurological diseases and injuries. The most important concept of the model is the assumption that every disease can be evaluated on four levels, namely: – – – – pathology level, impairment level, disability level, handicap level.

In practice, the border between individual levels can be quite blurred. In general, a distinction has to be made between measurements and scoring scales. Measurement involves the use of a specific, traceable standard and the comparison of an actual value of a certain quantity against this standard. Scales often rely on subjective observations or rough measurements leading to a very few quantum descriptors. In rehabilitation, measurements are much rarer than scoring scales. The simplest scoring of motor abilities and impairment of e.g. the upper extremity includes the squeezing of the dynamometer for estimating muscle strength and measuring the range of movement of individual joints. Self-scoring by patients is also present. The criteria of a good scoring scale include: – validity (the result actually refers to the aspects for which the test was carried out), – reliability (competent observer and time stability of results of such observer), – sensitivity (detection, differentiation of sufficiently small changes that are still relevant).

M. Munih and T. Bajd / Rehabilitation robotics

493

Standardised rehabilitation tests are for the most part tailored to a specific disease and consider also the condition of other parts of the body and activity in general. Some of these tests are Fugl-Meyr, Barthel Index and the widely-used Barthel Index, Katz index, Nottingham Adl Index, Jebsen, FIM-test, Box and Block Test, Nine-Hole Peg Test, Action Research Arm Test, Franchay Arm Test and others. Unlike the scoring approach, haptic, robotic technology provides for the next stage in rehabilitation, objective measurement during exercise. Quantities that vary with time (position and orientation, velocities, accelerations, forces, torques) can often be measured directly or derived. The presentation of these quantities for individual, relevant points of the body offers an objective basis for evaluation, as in kinesiology, and derived indexes can answer significant questions. Observation of measured data in a frequency domain can result in new insights, for instance, in the case of Parkinson’s disease, the amplitude and frequency of shaking are determined by both place and time. The measured quantities can be used as entries into various physical or physiological models and through them new important parameters may be acquired, e.g. active torques of the muscles of individual joints, passive torques in joints or even mass and evaluation of moment of inertia of some body segments. In doing so, we have to be aware of the fact that the more complex models usually comprise a higher number of parameters, some roughly evaluated, carrying high measurement uncertainly, which in the end may contribute to an even more uncertain final result. 8. Conclusion Recent technological developments have made many things possible for the first time. However, actual systems are only now being introduced into our lives in the fields of computing, virtual reality visualisation, medicine in general and a narrow field of rehabilitation. Previously most simple devices, accessories and physiotherapists’ hands will in the future be complemented by computerized devices. These will help in existing and, hopefully, also some new aspects of rehabilitation. Robotic approaches undoubtedly allow for equally fast, perhaps even faster, but certainly more interesting and entertaining rehabilitation methods. It has to be expected that at least some good laboratory prototypes will develop into widely-available products. These products are expected to have a suitable modular design, and developments in other areas, e.g. improvement in speed of computer processing, improvement in video techniques, software support, etc., will further contribute to the advancement in this area. Unfortunately, the size of the rehabilitation robotic market will be considerably smaller than the size of the market for computer games, which is why the volume of investment will be smaller and the pace of development is expected to be slower. Acknowledgements The described research work is financed by the Slovenian Research Agency in the form of a programme group Analysis and Synthesis of Movement in Man and Machine, and by providing grants to young researchers. References
[1] D. Khalili and M. Zomlefer, An intelligent robotic system for rehabilitation of joints and estimation of body segment parameters, IEEE Trans Biomed Eng 35 (1988), 138–146.

494 [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29]

M. Munih and T. Bajd / Rehabilitation robotics B. Gjelsvik and E. Bassoe, Form und Function, Stuttgart, New York: Thieme Verlag, 2002. J. Howle, Neuro-Developmental Treatment Approach: Theoretical Foundations and Principles of Practice, Laguna Beach, CA, NDTA, 2002. B. Bobath, Observations on adult hemiplegia and suggestions for treatment, Physiotherapy 45 (1959), 279–289. E. Panturin, The bobath concept, Clin Rehabil 15 (2001), 111–113. S. Raine, Defining the Bobath concept using the Delphi technique, Physiother Res Int 11 (2006), 4–13. F.A. Mussa-Ivaldi, N. Hogan and E. Bizzi, Neural, mechanical, and geometric factors subserving arm posture in humans, J Neurosci 5 (1985), 2732–2743. R. Shadmehr and F.A. Mussa-Ivaldi, Adaptive representation of dynamics during learning of a motor task, J Neurosci 14 (1994), 3208–3224. T. Tsuji, P.G. Morasso, K. Goto and K. Ito, Human hand impedance characteristics during maintained posture, Biol Cybern 72 (1995), 475–485. P. Lum, D. Reinkensmeyer, R. Mahoney, W.Z. Rymer and C. Burgar, Robotic devices for movement therapy after stroke: current status and challenges to clinical acceptance, Top Stroke Rehabil 8 (2002), 40–53. G. Fazekas, M. Horvath and A. Toth, A novel robot training system designed to supplement upper limb physiotherapy of patients with spastic hemiparesis, Int J Rehabil Res 29 (2006), 251–254. S. Hesse, G. Schulte-Tigges, M. Konrad, A. Bardeleben and C. Werner, Robot-assisted arm trainer for the passive and active practice of bilateral forearm and wrist movements in hemiparetic subjects, Arch Phys Med Rehabil 84 (2003) 915–920. C.D. Takahashi and D.J. Reinkensmeyer, Hemiparetic stroke impairs anticipatory control of arm movement, Exp Brain Res 149 (2003), 131–140. J.L. Patton and F.A. Mussa-Ivaldi, Robot-assisted adaptive training: custom force fields for teaching movement patterns, IEEE Trans Biomed Eng 51 (2004), 636–646. J.L. Patton, M.E. Stoykov, M. Kovic and F.A. Mussa-Ivaldi, Evaluation of robotic training forces that either enhance or reduce error in chronic hemiparetic stroke survivors, Exp Brain Res 168 (2006), 368–383. R.B. Salter et al., The effects of continuous compression on living articular cartilage, J Bone Joint Surg 42-A (1960). M.A. Ritter, V.S. Gandolf and K.S. Holston, Continuous passive motion versus physical therapy in total knee arthroplasty, Clin Orthop Relat Res 244 (1989), 239–243. H.I. Krebs, N. Hogan, B.T. Volpe, M.L. Aisen, L. Edelstein and C. Diels, Overview of clinical trials with MIT-MANUS: A robot-aided neuro-rehabilitation facility, Technol Health Care 7 (1999), 419–423. D.J. Reinkensmeyer, L.E. Kahn, M. Averbuch, A. McKenna-Cole, B.D. Schmit and W.Z. Rymer, Understanding and treating arm movement impairment after chronic brain injury: progress with the ARM guide, J Rehabil Res Dev 37 (2000), 653–662. C.G. Burgar, P.S. Lum, P.C. Shor and H.F. Machiel Van der Loos, Development of robots for rehabilitation therapy: the Palo Alto VA/Stanford experience, J Rehabil Res Dev 37 (2000), 663–673. S. Hesse, G. Schulte-Tigges, M. Konrad, A. Bardeleben and C. Werner, Robot-assisted arm trainer for the passive and active practice of bilateral forearm and wrist movements in hemiparetic subjects, Arch Phys Med Rehabil 84 (2003), 915–920. S. Coote, E. Stokes, B. Murphy and W. Harwin, The effect of GENTLE/s robot-mediated therapy on upper extremity dysfunction post stroke, Proc. 8th International Conference on Rehabilitation Robotics, 2003, Daejeon, Korea, HumanFriendly Welfare Robot System Engineering Research Center, 2003, pp. 59–61. C. Fanin, P. Gallina, A. Rossi, U. Zanatta and S. Masiero, NeRe-Bot: A wire-based robot for neurorehabilitation, Proc 8th International Conference on Rehabilitation Robotics, Daejeon, Korea, Human-Friendly Welfare Robot System Engineering Research Center, 2003, pp. 23–26. A. Toth, G. Fazekas, G. Arz, M. Jurak and M. Horvath, Passive robotic movement therapy of the spastic hemiparetic arm with REHAROB: Report of the first clinical test and the follow-up system improvement, Proc 9th International Conference on Rehabilitation Robotics, Chicago, Illinois. Madison (WI), Omnipress, 2005, pp. 127–130. J. Dewald, M.D. Ellis, B.G. Holubar, T. Sukal and A.M. Acosta, The robot application in the rehabilitation of stroke patients, Neurol Rehabil 4 (2004), 7. T. Nef and R. Riener, ARMin – Design of a novel arm rehabilitation robot, Proc. 9th International Conference on Rehabilitation Robotics, Chicago, Illinois. Madison (WI), Omnipress, 57–60, 2005 S.E. Fasoli, H.I. Krebs, J. Stein, W.R. Frontera and N. Hogan, Effects of robotic therapy on motor impairment and recovery in chronic stroke, Arch Phys Med Rehabil 84 (2003), 477–482. J. Stein, H.I. Krebs, W.R. Frontera, S.E. Fasoli, R. Hughes and N. Hogan, Comparison of two techniques of robot-aided upper limb exercise training after stroke, Am J Phys Med Rehabil 83 (2003), 720–728. R.J. Sanchez, E. Wolbrecht, R. Smith, J. Liu, S. Rao, S. Cramer, T. Rahman, J.E. Bobrow and D.J. Reinkensmeyer, A pneumatic robot for re-training arm movement after stroke: rationale and mechanical design, Proc 9th Int Conf Rehabilitation Robotics, Chicago, Illinois, Madison (WI), Omnipress, 2005, pp. 500–504.

M. Munih and T. Bajd / Rehabilitation robotics [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40] [41] [42] [43] [44] [45]

495

R.F. Beer, J.D. Given and J.P. Dewald, Task-dependent weakness at the elbow in patients with hemiparesis, Arch Phys Med Rehabil 80 (1999), 766–772. R.F. Beer, J.P. Dewald and W.Z. Ryme, Deficits in the coordination of multijoint arm movements in patients with hemiparesis: evidence for disturbed control of limb dynamics, Exp Brain Res 131 (2000), 305–319. R.F. Beer, J.P. Dewald, M.L. Dawson and W.Z. Rymer, Targetdependent differences between free and constrained arm movements in chronic hemiparesis, Exp Brain Res 156 (2004), 458–470. J. Furusho, K. Koyanagi, Y. Imada, Y. Fujii, K. Nakanishi, K. Domen, K. Miyakoshi, U. Ryu, S. Takenaka and A. Inoue, A 3-D rehabilitation system for upper limbs developed in a 5-year NEDO project and its clinical testing, Proc 9th Int Conf Rehabilitation Robotics, Chicago, Illinois. Madison (WI), Omnipress, 2005, pp. 53–56. R. Colombo, F. Pisano, S. Micera, A. Mazzone, C. Delconte, M.C. Carrozza, P. Dario and G. Minuco, Upper limb rehabilitation and evaluation of stroke patients using robot-aided techniques, Proc 9th Int Conf Rehabilitation Robotics, Chicago, Illinois, Madison (WI), Omnipress, 2005, pp. 515–518. L.E. Kahn, M. Averbuch, W.Z. Rymer and D.J. Reinkensmeyer, Comparison of robot-assisted reaching to free reaching in promoting recovery from chronic stroke, in: Integration of Assistive Technology in the Information Age, M. Mokhtari, ed., Amsterdam (the Netherlands), IOS Press, 2001, pp. 39–44. G.B. Prange, M.J.A. Jannink, C.G.M. Groothuis-Oudshoorn, H.J. Hermens and M.J. Ijzerman, Systematic review of the effect of robot-aided therapy on recovery of the hemiplegic arm after stroke, J Rehabilitation Research & Development 43 (2006), 171–184. S. Hesse and D. Uhlenbrock, A mechanized gait trainer for restoration of gait, J Rehabilitation Research & Development 37 (2000), 701–708. H. Schmidt, D. Sorowka, F. Piorko, N. Marhoul and R. Bernhardt, Entwicklung eines robotergest¨ tzten Laufsimulators u zur Gangrehabilitation (Development of a Robotic Walking Simulator for gait rehabilitation), Biomed Eng 40 (2003), 281–286. J.C. Perry and J. Rosen, Design of a 7-Degree-of-freedom upper limb powered exoskeleton, BioRob 2006, The IEEE/RASEMBS Int Conf Biomedical Robotics and Biomechatronics, Pisa, Toscany, Italy, 2006, pp. 805–810. H. Schmidt, D. Sorowka, F. Piorko, N. Marhoul and R. Bernhardt, Entwicklung eines robotergest¨ tzten Laufsimulators u zur Gangrehabilitation (Development of a robotic walking simulator for gait rehabilitation), Biomedical Engineering 40 (2003), 281–286. A. Frisoli, F. Rocchi, S. Marcheschi, A. Dettori, F. Salsedo and M. Bergamasco, A new force-feedback arm exoskeleton for haptic interaction in Virtual Environments, First Joint Eurohaptics Conference and Symposium on Haptic Interfaces for Virtual Environment and Teleoperator Systems WHC, 2005. M. Casadio, V. Sanguineti, P.G. Morasso and V. Arrichiello, Braccio di Ferro: A new haptic workstation for neuromotor rehabilitation, Technol Health Care 14 (2006), 123–142. K. Koyanagi, J. Furusho, U. Ryu and A. Inoue, Development of rehabilitation system for the upper limbs in a NEDO project, Proc 2003 IEEE Int Conf Robotics & Automation, Taipei, Taiwan (2003), 14–19. J.C. Perry, J. Rosen and S. Burns, Upper-limb powered exoskeleton design, IEEE Transactions on Mechatronics 12 (2007), 408–417. U. Mali and M. Munih, HIFE-haptic interface for finger exercise, IEEE/ASME Transactions on Mechatronics 11 (2006), 93–102.

Copyright of Technology & Health Care is the property of IOS Press and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.