Medicine Engineers, Part H: Journal of Engineering in ...

Proceedings of the Institution of Mechanical Engineers, Part H: Journal of Engineering in Medicine http://pih.sagepub.com/

Virtual Reality and a Haptic Master-Slave Set-Up in Post-Stroke Upper-Limb Rehabilitation J A Houtsma and F J A M Van Houten Proceedings of the Institution of Mechanical Engineers, Part H: Journal of Engineering in Medicine 2006 220: 715 DOI: 10.1243/09544119H06104 The online version of this article can be found at: http://pih.sagepub.com/content/220/6/715

Published by: http://www.sagepublications.com

On behalf of:

Institution of Mechanical Engineers

Additional services and information for Proceedings of the Institution of Mechanical Engineers, Part H: Journal of Engineering in Medicine can be found at: Email Alerts: http://pih.sagepub.com/cgi/alerts Subscriptions: http://pih.sagepub.com/subscriptions Reprints: http://www.sagepub.com/journalsReprints.nav Permissions: http://www.sagepub.com/journalsPermissions.nav Citations: http://pih.sagepub.com/content/220/6/715.refs.html

>> Version of Record - Jun 1, 2006 What is This?

Downloaded from pih.sagepub.com at Tongji University on September 21, 2011

TECHNICAL NOTE

715

Virtual reality and a haptic master–slave set-up in post-stroke upper-limb rehabilitation J A Houtsma and F J A M Van Houten Department of Engineering Technology, University of Twente, Enschede, The Netherlands The manuscript was received on 20 September 2004 and was accepted after revision for publication on 19 April 2006. DOI: 10.1243/09544119H06104

Abstract: Since the number of stroke patients is ever growing, and since studies show that the patients can benefit from more exercise, the possible use of virtual reality and haptic devices in the rehabilitation of stroke patients is researched. Besides offering more independent and challenging exercise possibilities, the inherent registration capabilities of the haptic devices open up the possibility of monitoring patients’ motions in an objective way. This research results in the implementation of a VR mock-up of an exercise in which patients use both hands and a stick to push away a ball that is tossed at them. In this mock-up, two FCS HapticMasters are used in a master–slave set-up to mimic the behaviour of the stick. The implementation of the scene shows that there are currently limitations, in the hardware as well as in the software, that cause the virtual exercise to behave differently to the real-life exercise. Consequences for the therapeutic value are subject to further research. Keywords: stroke, CVA, rehabilitation, virtual reality, haptic device, master–slave

1 INTRODUCTION

can be mimicked in a VR environment and discusses its shortcomings and possibilities.

Cerebral vascular accidents (CVAs), with often complete or partial single-sided paralysis as a result, are one of the main causes of disablement in Europe. Following a stroke, physiotherapeutic treatment is normally used to return the patient’s motor functions to the extent that this is possible. In post-stroke upper-limb treatment, an important task of the therapist is to guide repetitive movements, both physically and through verbal feedback. Different studies show that an increased amount of training is beneficial for the recovery of a patient. Several tools have been developed that offer the opportunity to train more independently [1–4]. A way of presenting patients with an independent, but controlled and lifelike environment is virtual reality (VR), which is also used in entertainment and engineering [5, 6]. This technical note shows a possible way of using VR, in combination with an admittance-controlled haptic device, in upper-limb stroke rehabilitation. It describes how a real exercise * Corresponding author: Department of Engineering Technology, University of Twente, PO Box 217, Enschede, 7500AE, The Netherlands. email: [email protected]

H06104 © IMechE 2006

2 DESIGN 2.1 Task selection A ball-hitting exercise was used as a basis for a VR mock-up. In this exercise (Fig. 1), the patient holds a stick horizontally in front of the body with two hands. A ball, tossed in the patient’s direction, has to be pushed away with the stick. The exercise has several goals. One is to improve the coordination and timing of the patient’s upper limb. The second is to enlarge the patient’s range of motion (ROM) by gradually enlarging the area in which the ball is thrown. The purpose of the horizontal stick is to present the patient with two therapeutically important kinds of feedback [4]. Its orientation provides visual feedback on whether or not the positions of both hands are identical. At the same time there is kinaesthetic feedback because the unaffected limb guides the affected one through the stick. This guidance is limited, however, because the unaffected wrist often Proc. IMechE Vol. 220 Part H: J. Engineering in Medicine


716

J A Houtsma and F J A M Van Houten

Fig. 1 Still of a patient performing an exercise in which tossed balls are pushed with a stick

cannot produce enough torque to lift the affected arm. When properly mimicked, the inherent feedback of the exercise, and the many possible difficulty settings (e.g. ball speed and size), provide excellent opportunities for a therapist-independent VR exercise. 2.2 Visual display For the visual representation of the scene, a screen of 1.10 m×1.47 m is used with a resolution of 600×800 pixels. The scene is presented in monoview, meaning there is no real perception of depth. A screen shot of the scene (Fig. 2) shows a solid floor, the stick that follows the user’s movements, and the ball that interacts with the other objects in the scene. This visualization is done with open-source, OpenGL-based software called ‘Draw stuff’ and uses perspective and shadow to offer a sense of threedimensional view. 2.3 Haptic set-up In the set-up, the kinaesthetic part of the exercise is represented by two FCS HapticMasters (Fig. 3) in a

Fig. 3 The workspace of the HapticMaster

master–slave configuration. These haptic devices measure 1.00 m×0.35 m×0.91 m (length×width× height) and have three driven degrees of freedom (DOF): two translations and 1 rotation [7]. They are controlled through a dedicated server with an update frequency of 2500 Hz. There are several reasons for using the HapticMaster. First, with a ROM of 0.36 m×0.27–0.61 m× 0.40 m (depth × width × height), it approximately fits the ROM of the average patient. Also, with a nominal force output of 100 N (250 N peak), it can exert enough force to properly guide or even lift a limb of about 3–5 kg. Furthermore, the HapticMaster uses an admittance control scheme instead of an impedance control scheme, like most other haptics. Impedance control measures the displacement of the haptic and gives corresponding force feedback. Admittance control takes the exerted force, measured by a force sensor, as the major input and puts out the corresponding displacement. This paradigm makes it possible to compensate for the mass and friction of the devices, exert large forces, and handle well in master–slave applications. The force sensor that is used also provides an opportunity to objectively measure the interaction forces while registering the ROM of the user. Figure 4 shows a subject using the system. For patients, the ball-shaped end-effectors can be replaced by a (rotating) splint, as done in the GENTLE/s project [8]. 2.4 Process control

Fig. 2 Screen shot of the ball hitting exercise in VR

In order for the VR exercise to be controlled, a program was developed that runs on a host PC (dual

Proc. IMechE Vol. 220 Part H: J. Engineering in Medicine


H06104 © IMechE 2006

Post-stroke upper-limb rehabilitation

Fig. 4 A subject performing the ball-hitting exercise in VR by controlling two HapticMasters

Athlon MP 2000+ processors, 1024 MB internal memory, ATI 9700 PRO 128 MB graphics card). Figure 5 shows a schematic representation of the control loop, which continually runs through the following steps. 1. The interaction forces are calculated. Because the exercise uses several moving objects, these cannot be created in the static environment of the dedicated server. Therefore, an open-source, multirigid-body-dynamics engine called ODE [9] is used to calculate the interactions and, with that, the haptic feedback of the simulation. 2. The master–slave forces are calculated in a separate routine. This connects the positions of both

717

haptics by using a linear spring–damper equation, which calls both haptics to obtain the forces exerted by the user. 3. The calculated forces are added and sent to the HapticMasters, where they are compared with the exerted forces. The resulting position and velocity are returned to the loop. 4. The position is used both to update the visual rendering of the scene and to calculate the dynamic interaction forces in the next cycle. This implies that the visual update frequency is equal to the haptic update frequency. 5. While the main loop is executed, the keyboard is monitored for input.

3 PERFORMANCE In general, the visuals look smooth, with realistic interactions of objects. The lack of stereoscopy does complicate the coordination within the scene. The use of shadow, on the other hand, improves the perception of distance between the objects, as does moving the stick around for better reference. Although the haptic representation works satisfactorily when rendering smaller forces, large forces create instabilities. This makes it hard to guide an entire arm, as intended. Furthermore, the haptic simulation sometimes suffers from glitches, which express themselves in sudden loss or increase of force feedback. The control loop runs at a frequency of around 380 Hz, which determines both the visual and the haptic performance. This frequency is dominated by the four calls of 0.6 ms each that are made to the haptics every cycle with the master–slave routine in use.

4 EVALUATION 4.1 Design evaluation

Fig. 5 Visual representation of the control loop of the program H06104 © IMechE 2006

The visual representation of the exercise did not present many problems. In order to improve coordination, however, stereoscopy should be implemented. With the current software and shutter glasses or a head-mounted display (HMD), this should be relatively easy. The use of multi-rigid-body dynamics to compute haptic interaction appears to work well, and allows for flexible environments. The control loop frequency is 380 Hz, while a minimum of 1000 Hz is recommended for haptic rendering of large forces. With the new HapticAPI, fewer Proc. IMechE Vol. 220 Part H: J. Engineering in Medicine


718

J A Houtsma and F J A M Van Houten

communication commands are needed for the same functionality, allowing for a substantially higher update frequency. Since the visual update frequency for stereoscopy only needs to be 120 Hz, visual and haptic rendering should also be split into separate threads, running simultaneously at different speeds. This will raise the haptic loop speed, improving the scene stability.

perception needed for proper optokinetic coordination. 4. The therapeutic value of the current system should be subject to further research. If necessary, the system should be expanded with more sensors and devices to guide the movement so it is more adequate for independent training (master–slave configuration) as well as for progress evaluation (independent configuration).

4.2 Therapeutic effectiveness With proper visual equipment (HMD), the produced scene and set-up can probably be used to train the coordination, timing, and ROM, as intended. Added effects like damping can even be used to stabilize motions. The therapeutic effectiveness largely depends on the quality of the feedback. Compared with the original one, this exercise provides less visual feedback, because the coupled positions of both haptics result in a fixed orientation of the virtual stick. The kinaesthetic feedback, however, is larger because the affected hand can be guided much better. Although the inherent registering possibilities of the HapticMaster are much more exact and objective than a therapist can be, this system only provides feedback on the position and force of the hand, instead of the entire limb. This limits the therapeutic possibilities of the current set-up.

5 CONCLUSIONS 1. A master–slave arrangement of two haptic devices appears to be a powerful therapeutic tool for stroke victims to regain coordination and control over an affected upper limb through mechanical feedback, indirectly provided through their other unaffected limb. 2. To avoid disturbing haptic artefacts caused by undersampling, the haptic update rate must be raised significantly. By using separate visual and haptic program loops and newer software, this should be quite feasible. 3. The visual image representation should also be made stereoscopic to give the patient the depth

REFERENCES 1 Volpe, B. T., Krebs, H. I., and Hogan, N. Is robotaided sensorimotor training in stroke rehabilitation a realistic option? Current Opinion in Neurology, 2001, 14(6), 745–752. 2 Lum, P. S., Burgar, C. G., Shor, P. C., Majmundar, M., and Van der Loos, M. Robot-assisted movement training compared with conventional therapy techniques for the rehabilitation of upper-limb motor function after stroke. Arch. Phys. Med. Rehabil., 2002, 83(7), 952–959. 3 Boian, R., Sharma, A., Han, C., Merians, A. S., and Burdea, G. Virtual reality-based post-stroke hand rehabilitation. In Proceedings of Medicine Meets Virtual Reality Conference 2002, Newport Beach, California, 2002, pp. 64–70 (IOS Press, Amsterdam). 4 Reinkensmeyer, D. J., Takahashi, C. D., Timoszyk, W. K., Reinkensmeyer, A. N., and Kahn, L. E. Design of robot assistance for arm movement therapy following stroke. Advd Robotics, 2000, 14(7), 625–637. 5 Van Houten, F. J. A. M. The use and development of haptic devices and virtual reality as engineering tools. CIRP J. Mfg Syst., 2003, 32(5), 387–397. 6 Tideman, M., Van der Voort, M. C., and Van Houten, F. J. A. M. A haptic virtual prototyping environment for design and assessment of gearshifting behavior. In Proceedings of the 14th International CIRP Design Seminar, Cairo, Egypt, 2004. 7 Van der Linde, R. Q., Lammertse, P., Frederiksen, E., and Ruiter, B. The HapticMaster, a new highperformance haptic interface. Proceedings of Eurohaptics 2002 Conference, University of Edinburgh, Edinburgh, UK, 2002, pp. 1–5. 8 Loureiro, R. and Amirabdollahian, F. Upper limb robot mediated stroke therapy – GENTLE/s approach. Autonomous Robots, 2003, 15, 35–51. 9 Website open dynamics engine. Available from http://ode.org/ [accessed 12 September 2004].

Proc. IMechE Vol. 220 Part H: J. Engineering in Medicine


H06104 © IMechE 2006