Fraunhofer-Institut für Produktionsanlagen und Konstruktionstechnik IPK

Filters

13
Reset filters

Settings
Now showing 1 - 10 of 13
  • Publication
    Patient adaptive control of end-effector based gait rehabilitation devices using a haptic control framework
    ( 2011)
    Hussein, S.
    ;
    Krüger, J.
    Robot assisted training has proven beneficial as an extension of conventional therapy to improve rehabilitation outcome. Further facilitation of this positive impact is expected from the application of cooperative control algorithms to increase the patient's contribution to the training effort according to his level of ability. This paper presents an approach for cooperative training for end-effector based gait rehabilitation devices. Thereby it provides the basis to firstly establish sophisticated cooperative control methods in this class of devices. It uses a haptic control framework to synthesize and render complex, task specific training environments, which are composed of polygonal primitives. Training assistance is integrated as part of the environment into the haptic control framework. A compliant window is moved along a nominal training trajectory compliantly guiding and supporting the foot motion. The level of assistance is adjusted via the stiffness of the moving window. Further an iterative learning algorithm is used to automatically adjust this assistance level. Stable haptic rendering of the dynamic training environments and adaptive movement assistance have been evaluated in two example training scenarios: treadmill walking and stair climbing. Data from preliminary trials with one healthy subject is provided in this paper.
  • Publication
    Visuelles Biofeedback für die gerätegestützte neurologische Gangrehabilitation nach Schlaganfall
    ( 2010)
    Brüning, M.
    ;
    Hussein, S.
    ;
    Bardeleben, A.
    ;
    Schmidt, H.
    ;
    Krüger, J.
    ;
    Hesse, S.
  • Publication
    Richtlinien für die Gestaltung von visuellem Biofeedback in der neurologischen Gangrehabilitation nach Schlaganfall
    ( 2010)
    Brüning, M.
    ;
    Hussein, S.
    ;
    Bardeleben, A.
    ;
    Schmidt, H.
    ;
    Krüger, J.
    ;
    Hesse, S.
  • Publication
    Structural analysis method for optimized design of complex kinematic structures using static and dynamic models and application to a robotic walking simulator
    ( 2010)
    Brüning, M.
    ;
    Hussein, S.
    ;
    Schmidt, H.
    ;
    Krüger, J.
    This paper describes an approach for structural analysis for design improvements of complex, e.g. hybrid, kinematic structures utilizing static and dynamic models. It is suitable to locate improvement potentials in existing mechanisms, facilitate goal-oriented design of new mechanisms or for a simulation-based controller synthesis e.g. a compliance-controller. To receive a model close to reality, mechanical influences, which are commonly neglected in conventional robot models, are analyzed regarding their relevance and if suitable integrated into the model. Investigated effects are the mechanical compliances of links and gears, compliances of the actuators resulting from the control circuits as well as non-linear frictional influences of the actuators. The kinematic and dynamic model is realized as an iterative solution instead of a closed analytic solution with extensive symbolic expressions. This leads to an analysis with clearly arranged aspects, further more the model is suitable for usage in a real-time control. The mechanical influences are analyzed analytically. The derived dynamic modeling is based on the Newton-Euler formulation. The approach is applied to the robotic walking simulator HapticWalker, a device for robot assisted gait rehabilitation. It consists of two identical hybrid parallel-serial manipulators. The forces calculated by the use of the developed model are in a good congruence with measured values. An obviously improved correspondence between measured and calculated values is achieved by the non-linear friction model of the actuators.
  • Publication
    Effect of different training modes on ground reaction forces during robot assisted floor walking and stair climbing
    ( 2009)
    Hussein, S.
    ;
    Schmidt, H.
    ;
    Hesse, S.
    ;
    Krüger, J.
    The HapticWalker is a robotic walking simulator for neurological gait rehabilitation. The device is based on the end-effector principle, i.e. the patients' feet are fixed on two footplates, his trunk is secured via a suspension harness. The footplates (end-effectors) of the robot guide the feet of the patient on freely programmable trajectories. Currently two different training modes are provided for each foot trajectory on the HapticWalker. Firstly training trajectories based on real motion capture data can be used. Secondly the vertical center of mass (CoM) motion can be incorporated into the foot trajectory. The latter method leaves the absolute position of the CoM constant, while the relative movement between CoM and feet remains the same as in free walking. Thereby the v ertical CoM motion can be actively supported by using a fixed passive suspension, any type of controlled body weight support will then be accomplished via force controlled footplates. To characterize the training on the machine in position controlled mode, muscle activities as well as ground reaction force data of healthy subjects were collected. This article presents preliminary results from the investigation of ground reaction force data of healthy subjects. Therefore data of 10 healthy subjects was used to compare HapticWalker training modes vs. free walking on level ground and upstairs without any assistive devices and the aforementioned two HapticWalker training modes against each other. Amplitude and timing of force peaks during heel strike, foot flat and push off in vertical ground reaction forces were used for the assessment. In spite of a footplate, which does not provide a separate metatarsal joint for relative motion between fore and hind foot, during heel strike and push off the known characteristic peaks at the beginning and the end of the stance phase can be clearly distinguished. Though compared to free walking a slower and premature rise and slower decline of ground reaction for- ces can be seen due to the stiff guidance of the feet in position controlled mode. No significant differences appeared between the two described HapticWalker training modes.
  • Publication
    Multimodal biofeedback architecture for an electromechanical gait rehabilitation device
    ( 2009)
    Hussein, S.
    ;
    Schmidt, H.
    ;
    Hesse, S.
    ;
    Krüger, J.
  • Publication
    Adaptive control of an end-effector based electromechanical gait rehabilitation device
    ( 2009)
    Hussein, S.
    ;
    Schmidt, H.
    ;
    Krüger, J.
    In industrialized countries stroke is the major cause for physical disabilities in adults. In various clinical studies gait therapy with the help of the electromechanical Gait Trainer GT-I proved to enhance the rehabilitation outcome for subacute stroke patients. This paper presents control methods that were developed to enable variability during treatment in order to further improve gait therapy with this class of devices. The algorithms suitable for the Gait Trainer GT-I are analyzed in a simulation study. Therefore models which simulate the practicing subjects behaviour were developed. A purely mechanical mass-damper system models the passive subjects behaviour while motor learning models were adopted to simulate patient adaptation different types of footplate gui dance characteristics. Several adaptive approaches have been developed for other rehabilitation devices in the past. In this work two controllers were developed and evaluated. The first features a one dimensional control window along the footplate trajectory within which the patient is only slightly guided. Outside the window a force field draws the subject back to the window. The second algorithm extends the window controller with a human motor learning strategy for to adapt the window size and thereby the assistance provided to the subjects. They were tested in a simulation study with different human behaviour models, the results are presented in this paper.
  • Publication
    Ground reaction forces during different floor walking in robot assisted gait training
    ( 2009)
    Hussein, S.
    ;
    Schmidt, H.
    ;
    Hesse, S.
    ;
    Krüger, J.
  • Publication
    Muscle activation of stroke patients during stair climbing in robot assisted gait training
    ( 2008)
    Hussein, S.
    ;
    Schmidt, H.
    ;
    Volkmar, M.
    ;
    Krüger, J.
    ;
    Hesse, S.
    For central nervous system (CNS) impaired patients, e.g. after stroke, the achievement of an independent, stable gait function and endurance in walking is essential for independent mobility in daily life. Modern concepts of rehabilitation favor a task specific repetitive training, that facilitates natural motion and muscle activation patterns, i.e. in addition to commonly practiced floor walking other gait motions of daily living (ADL), such as stair climbing, should be trained. In gait rehabilitation of severely affected non-ambulatory patients, a guided training must be applied as no or only little voluntary motor control is left in the affected side. The rehabilitation robot HapticWalker is the first device that allows for robot assisted guided training of arbitrary foot trajectories, e.g. also stair climbing. The goal of this study was to assess the effect of guided training in stair climbing condition on the HapticWalker on the movement and muscle activation patterns of stroke patients. Generally rhythmic and phasic muscle activation patterns were observed, which is an important factor in gait rehabilitation. The training of stair climbing condition on the HapticWalker also facilitated body weight shift in all patients and proper activation of weight bearing muscles. The HapticWalker currently lacks a lateral hip guidance facility, hence individual differences in the ability of controlling hip movement and body weight shift depending on the level of gait ability were observed.