Jtam-A4.dvi JOURNAL OF THEORETICAL AND APPLIED MECHANICS 56, 4, pp. 1179-1191, Warsaw 2018 DOI: 10.15632/jtam-pl.56.4.1179 COMPUTATIONAL OPTIMIZATION AND IMPLEMENTATION OF CONTROL SYSTEM FOR MECHATRONIC TREADMILL WITH BODY WEIGHT SUPPORT SYSTEM Grzegorz Gembalczyk, Sławomir Duda, Eugeniusz Świtoński Silesian University of Technology, Department of Theoretical and Applied Mechanics, Gliwice, Poland e-mail: grzegorz.gembalczyk@polsl.pl; slawomir.duda@polsl.pl; eugeniusz.switonski@polsl.pl The purpose of this paper is to present a novel mechatronic system for gait re-education which consists of a body weight support system (BWS system) and a treadmill. This pu- blication covers mainly issues related to the design and optimization process of a control algorithm dedicated for the unloading system. The proposed control system is based on a fuzzy logic controller coupled with a PID regulator. The optimization of parameters for regulators has been conducted based on numerical simulations in which a hybrid optimiza- tion method combining a genetic algorithm with a gradient algorithm has been used. The developed control system has been tested experimentally. Keywords: gait reeducation, control system, optimization, fuzzy logic 1. Introduction Half of all deaths are caused by cardiovascular diseases in themodernworld. It is estimated that thenumberof deathsdue to stroke can reach7.69mlnpeople to 2030, and itwill constitute oneof more important medical problems. Strokes constitute the main reason for permanent disability in the adult population, particularly after turning 60. They are one of the main reasons for remaining in the condition of disability for a longer period of time. The restoration of possibility to move or keep balance in the upright position is significant for patient’s proper functioning. The deficit connected with the lack of ability to walk pertains to almost 70% of patients who have experienced a stroke. Kinesiotherapy is the most common and obvious way of treatment after a stroke that shouldbe started as soon as possible.The simplest formof exercises iswalking in company of a physiotherapist using simple devices for support such as walkers or banisters. Mechatronic deviceswhich support the process of kinesiotherapy after a stroke are in some sense “packed with the latest technological solutions”. They affect training efficiency and, especially, improve the speed of walking, endurance and balance. Such devices support physiotherapists relieving them fromhardphysicalwork of supportingapatient andallowing themto concentrate on the essence of the exercises carried out to a greater degree (Koceska andKoceski, 2013; Kot and Nawrocka, 2012). These devices both implement the rehabilitation process and verify the progress of treatment. When analyzing rehabilitation devices for kinesiotherapy within a dozen years, it is possible to notice that unloading a patient is the most significant element subject to continuous modifications. It is possible to find in the market devices with a simple design that make it possible to unload a patient in a passive way as well as technologically advanced solutions with active body weight support systems. In the case of passive systems (unloading force regulated by a crank, counterweight, spring, etc.), the difference between the unloading force that is set and the one that ismeasured during exercises ismuch bigger (Frey et al., 2006). Therefore, the exercises with devices equipped in body weight support systems with active control of the unloading force are more beneficial for patients. When comparing rehabilitation 1180 G. Gembalczyk et al. devices equipped with active body weight support systems with one another, it is possible to notice that the unloading mechanism is an element connected with a still supporting structure (it does not keep upwith a patient’s movement) inmost cases; and the patient’s walk forces the manipulator (mechanic orthosis) or treadmill movement (Chen et al., 2013; Pajor and Herbin, 2015). In the second case, training is recommended for people who have regained the ability to walk in a degree that makes it possible to move independently, but they still require support and improvement of the technique, speed and stamina (Mehrholz et al., 2014; Cao et al., 2014; Querry et al., 2008). The technologies of biofeedback and virtual reality are also used in the devices of this type (Koenig et al., 2011; Lünenburger et al., 2007, Jurkojć et al., 2017). A device in which the movement of a training person is limited only by the space of a supporting frame, onwhich aBWS systemmoves, has been created as a result of research at the Institute of Theoretical and Applied Mechanics financed by the National Centre for Research andDevelopment (Duda et al., 2016). The possibility of free sidemovements of a patient’s pelvis plays a significant role in rehabilitation exercises that restore the proper pattern of walking, as research shows (Dragunas and Gordon, 2016; Mignardot et al., 2017). A new form of a mechatronic device for the re-education of walking has been currently developed. It is a combination of amovable bodyweight support systemwith a treadmill. The thing that differentiates the proposed solution from those existing on themarket results from coupling the movement of an exercising personwith themovement of a treadmill tape. It is not the treadmill movement that forces patient’s movement. It is the patient that forces treadmill movement. It is implemented by using a sensor measuring the inclination angle of a sling rope. Moreover, the device allows free side movements that are so important for proper rehabilitation. It is achieved thanks todrive in adirection that is transverse to thedirection ofwalking.The invented mechatronic system for gait re-education includes four drives: • two independently working drives of the bodyweight support system, • the drive of awinch trolleymaking it possible for training aperson tomove in the direction that is perpendicular to the plane of the patient’s movement, • the drive of the treadmill tape. This publication involves an attempt at developing and optimizing a control system of the body weight support system with a fuzzy logic controller. No such solutions have been found in literature. Systems used for unloading a patient with an automatically regulated force are usually described in a very general way.Most of all, there are no descriptions of systems steering winches equipped with two drives cooperating with each other. The minimization of mass and size constitutes an important construction condition in the case of unloading systemswithmobile BWSsystems thathavebeen suspendedto the ceiling in the current solutions (Hidler et al., 2011; Duda et al., 2016).However, it is connectedwith the limitation of a linear drivemovement range. The right synchronization of the work of a winding drum with a linear actuator is necessary if one wants to make it possible for patients to do exercises with significant movements. Due to limited movement of such a linear actuator, the development of an optimal steering system constitutes an interesting engineering problem. This is why the problem of optimization of a system controlling the BWS system has been tackled in this article. Themost detailed description of a similar systemwas presented by Frey et al. (2006) descri- bing theLokolift device. Awindingdrumcooperateswith a rope (passive) drive in that solution. In order to steer this winding drum, switches have been usedwhich are activated at themoment when themovement of the linear actuator (instantaneous or mean) goes beyond the admissible range. Thus, the engine that propels the rope drum works only with the nominal speed or it does not work at all. The Lokolift system is meant mainly for the cooperation with a patient training on the treadmill, and its whole construction is placed on the floor. Therefore, this device does not have Computational optimization and implementation of control system... 1181 to meet restrictive criteria connected with the minimization of mass and dimensions, so the range of the linear actuator is not limited by any restrictive conditions. Taking into account the way devices for the re-education of walking develop and the beneficial influence of mobile body weight support systems on the stability of a patient’s walk when training in reduced weight conditions, it is possible to expect further development of similar devices (Reinkensmeyer and Dietz, 2016; Riener et al., 2010). 2. Description of the device The device used for the re-education of walking presented in this paper is amechatronic system in which the bodyweight support system, the system of compensating side inclinations and the training treadmill have been integrated (Fig. 1). Fig. 1. Mechatronic device for gait re-education A training person is pinned up in a special orthopaedic harness fastened to the rope of a winchingdevice.Thecouplingbetween thementioned subsystemsandthepatient is implemented throughmeasurement systems, onemeasuring theunloading force andtheothermeasuring linear movements resulting fromthedeflection of the line fromthevertical axis (sensors) andparticular mechanismdrives (actuators). The encoders of servodrives are additional servers that the device is equipped with. The transmission of measurement and control signals of the device takes place in the real timemode (Miądlicki andPajor, 2015; Sapiński et al., 2016)with theuse of a computer equipped with 2 RT-DAC4/PCI cards andMATLAB/Simulink software. The calculation step of the real time isT0 =0.01s.The servodrives of thewinchingdevice and compensation of side inclinations work in the speed control mode. Then, the generated steering signals retain the set rotational speed in a precise way. Steering the treadmill, described in details in (Duda et al., 2017), takes place by a modified original control panel. It has been achieved by parallel connection of an additional controller whose task involves simulation of the work of buttons placed on the panel that steers the treadmill. The electronic system has been built with the use of optocouplers steeredwith digital signals generated by the steering system in a PCComputer. The change of a signal from a high status to a low one and back to the high status has been obtained after 0.1s. It is equivalent to the change of the treadmill speed obtained after pressing the button (faster or slowly) on 1182 G. Gembalczyk et al. the control panel of the treadmill. This is also the reason for a relatively low sensitivity of the steering system to changes of walking parameters. A tracking systemwith a PD controller has been implemented for controlling themovement of the BWS system trolley (servodrive Y ) along a girder. Descriptions of such a system were presented in (Duda et al., 2016; Raczka et al., 2013). Controlling aBWS system requires the use of a more advanced active control systemwhich is presented in Section 5. 3. Modelling the body weight support system The processes of optimization of complex steering algorithms in which several or dozen parame- ters shouldbe specified are usually carried outwith offlinemethods.Thesemethods are basically necessary in the case of tuning steering systemswith devicesmeant for cooperationwith people; for rehabilitation in particular. In order to carry out the optimization of a system controlling a BWS system it is necessary to develop a numerical model taking into account the coupling between the electric part, themechanical one, and the patient. Amodel of a synchronousmotor with permanent PMSMmagnets (PermanentMagnet SynchronousMotor) described in (Mężyk et al., 2016; Xu, 2012) has been used for building the numerical model of the device. The body weight support system that is responsible for unloading the patient with the set force is themost important subsystemof thedevice presented in the article. In contrast to classic hoist devices, its driving system has been equipped with two independently working engines. The first one that is marked as Z1 is responsible for propelling the rope drum; the second one (Z2 linear actuator) is connected with the system of dynamic compensation. This system works on the drive basis of a Series Elastic Actuator type (SEA) (Pratt and Williamson, 1995; Robinson et al., 1999). Figure 2 presents a detailed construction of the driving system. The Fig. 2. The structure of the driving system of the body weight support system construction of the BWS system has been developed on the basis of aluminiumprofiles creating supporting frame (1) to which the following have been fixed: track guides (2), compression springs (3), driving engine of the Z2 axis, winding drum with an engine (5), and a system of pulleys. The construction of the presented system makes it possible to change rope length (6) through a change of the location of the pulley (7). This pulley is fixed to the first steel trolley out of two (8) that move along tracking guides (2) and are separated with springs (3). This diminishes the consequences of impact loads appearing during operation. Themovement of the second trolley (9) is implemented with a linear drive with ball screw (10) of a 5mmpitch whose rotation causes movement of a nut that is fastened to this trolley (9). The maximum range of theZ2 linear actuator movement is ca. 110mm. Computational optimization and implementation of control system... 1183 The developed physical model of the BWS system is shown in Fig. 3a, after including sim- plifying assumptions. Fig. 3. (a) The physical model of the BWS system. (b) The forces which act on the actuatorZ2 Three generalized coordinates connected with the rotation angles of rotators in the engines and the shift of the patient’s centre of gravity have been accepted in the mathematical model of the presented system q= [ϕZ1,ϕZ2,zp] (3.1) If one takes into account the fact that the resistance to movement and the trolley weight whose movement ismarkedwith thexc coordinate are negligibly small, then one can estimate the value of the unloading force influencing the patient (rope tension force) FUNL, when considering the forces influencing this trolley (Fig. 3b) FUNL ≈ Fs 2 Fs = ks ( ϕZ2 h 2π −xc ) + bs ( ϕ̇Z2 h 2 − ẋc ) (3.2) where xc = zp−ϕZ1iZ1rZ1 2 (3.3) In the equation above, iZ1 specifies the planetary gear ratio between the rope drum and theZ1 engine. In the mathematical model of the BWS system, the mass of the propelling screw and the trolley propelled by it have been reduced to the moment of inertia of the Z2 engine shaft. The inertia of other pulleys, the winding drum, and the planetary gear has been reduced to the moment of inertia of theZ1 engine shaft. In such a case, the searched equations of motion have the following form ϕ̈Z1 = 1 IZ1 (MZ1− bZ1ϕ̇Z1−FUNLiZ1rZ1) ϕ̈Z2 = 1 IZ2 ( MZ2− bZ2ϕ̇Z2−2FUNL h 2π ) z̈p = 1 mp (FUNL−mpg) (3.4) The coupling between the electric and mechanical part is implemented in such a way that the drivingmoment determined from the numericalmodel of the PMSMmotor (MZ1 andMZ2) has been implemented in themodel of themechanical part (3.4). The calculated angular acceleration (ϕ̈Z1 and ϕ̈Z2) has been, in turn, used in the engine model for calculating the rotation speed and the rotor rotation angle. 1184 G. Gembalczyk et al. Fig. 4. A functional scheme of a mechatronic treadmill for locomotive training Figure 4 shows a functional plan of the device in a block form in which the servo drives, themechanical part (BWS system and treadmill), patient’s movement, and the steering system have been included. Parameters of the mathematical model have been determined on the basis of catalog data, experimental tests on theuniversal testingmachineand in theestimationprocess.Theestimation of parameters in the models of Z1 and Z2 servo drives has been carried out on the basis of catalogue data and experimental data in which the rotational speed of the engine in various strain conditions has been registered.Thevalidation of themodel developed for theBWSsystem (Fig. 3a) has been carried out with the use of a weight of 20kg hung at the end of the rope. During the experiment, the rotational speed of 1000 (rotations/per minute) has been set on the Z1 engine; and the value of the unloading force has beenmeasured at the same time. Analogical tests have been carried out with the use of the Z2 drive. The correlation coefficient has been calculated with R-Pearson’s method (Gniłka andMężyk, 2015), and it equals 0.87. 4. Modelling the patient’s gait A kinematic input has been used in further numerical research for simulation of the patient’s movements. Thus, the third equation has been reduced in the system of equations of the BWS system movement, see Eq. (3.4). The equations describing patient’s movements zp [m] (used to calculate the forceF in accor- dance with equations (3.2) and (3.3)) have been described with mathematical functions whose formulas have been formulated on the basis of parameter measurements of walking of the disa- bled person. The results of the research of walk kinematics with participation of people after a stroke with paresis of the right side have been used in the model. The analysis of the walk kinematics has been carried out with the use of APAS software. Taking into account the fact that the BWS system should also guarantee a reliable work during doing other exercises in which vertical movements have a wide range (e.g. squats, climbing stairs), this input has been described by the system of equations zp =            0 for t< 2 0.016[sin(2π(t−2))− sin(4π(t−2))] for 2¬ t< 5.166 0.2(t−5.166) for 5.166¬ t< 6.166 0.2 for t­ 6.166 (4.1) Computational optimization and implementation of control system... 1185 5. Optimization of the control system The results of a preliminary experimental study of the control systems of the presented body weight support system has shown that the use of the Z2 drive only is sufficient for ordinary walking on a treadmill (Duda and Gembalczyk, 2016). In the case when a rehabilitated person is supposed to doadifferent exercisewhich involves a significant change of the trunk (e.g. squats, standing up from a chair, climbing obstacles), it is necessary to start a winding drum- the Z1 drive. In order to take advantage of the device potential fully and to guarantee a patient the biggest comfort of exercises done possible, one should develop and tune steering systems designated separately for every performed exercise or several main modes of the device operation. From the point of view of physiotherapists, a simple system with few options is desired. Therefore, a universal steering system that will be able to steer a device independently of the patient’s behaviour is anoptimal solution for them.Anattemptatdevelopingauniversal steering system of the BWS system has been made to meet these expectations taking into account economic aspects at the same time (energy efficiency). On the basis of earlier experience, the proposed steering systemhasbeenbased on aPIDcontroller coordinatingwith a fuzzy controller (Zhao et al., 2007). TheZ2 drive is steered in a feedback loop with the sensor of the unloading force, whereas theZ1 drive controller additionally usesmovementmeasurement of theZ2 drive actuator. This solution is presented in the form of a scheme in Fig. 5. Fig. 5. Block diagram of the proposed control system The basis of rules of the fuzzy controller has been prepared in such as way so that the Z1 drive is started only in the situation when the Z2 actuator moves away from the admissible working range or when the deviation of the reliving force is bigger than the accepted admissible deviation. The accepted set of rules is presented in the table. Sevenmembership functions have been defined for every input and output signal of the fuzzy controller. The symbol of Z2 disp means movement of the actuator in theZ2 drive, whereas ∆UNL is the value of the unloading force error ∆UNL(t)=Fref −FUNL(t) (5.1) The signs from “−−−” to “+++” represent the value from the lowest to the highest one. The “nom” indexmeans that the variable value is placed within the accepted admissible range, where the operation of Z1 is not required. The assumed admissible range for moving the Z2 actuator is 10mm(Z2Dnom =±10mm),andtheadmissibledeviation of theunloading force20N (∆UNLnom =±20N). Moreover, the following ranges of the input signals have been accepted ∆UNL∈ 〈−100;100〉 [N] Z2disp ∈ 〈−50;50〉 [mm] (5.2) 1186 G. Gembalczyk et al. Table 1.The basis of fuzzy logic controller rules Displacement of theZ2 actuator (Z2disp) −−− −− − nom + ++ +++ U n lo a d in g fo rc e ∆UNL−−− 0 0 Z1+ Z1++ Z1+++ Z1+++ Z1+++ ∆UNL−− 0 0 0 Z1+ Z1++ Z1+++ Z1+++ ∆UNL− 0 0 0 0 Z1+ Z1++ Z1+++ ∆UNLnom Z1−− Z1−− Z1− 0 Z+ Z1++ Z1++ ∆UNL+ Z1−−− Z1−− Z1− 0 0 0 0 ∆UNL++ Z1−−− Z1−−− Z1−− Z1− 0 0 0 ∆UNL+++ Z1−−− Z1−−− Z1−−− Z1−− Z1− 0 0 The optimal values of the PID controller settings and the range of the accepted member- ship functions of input signals in the fuzzy controller have been searched for in the process of optimization. Both controllers have been optimized separately due to the accepted assumptions. The input has been limited only to simulation of walking itself during the selection of the PID controller settings. The simulation time has been limited to t< 5.166 in Eq. (4.1). The settings of the PID controller have been the optimized variables, whereas the goal function has been connected with the minimization of the unloading force error FCPID = 5.166 ∫ 0 |∆UNL(t)| dt→min (5.3) The optimization of the fuzzy controller has been started with the acceptance of the rele- vant convention of generating membership functions. The coordinates of four additional points P1-P4 are calculated for each input in the proposed approach. The relevant membership func- tions are created in the next step on their basis. The values of these points are calculated with the use of the factors of proportionality included in the vector of the decision variables x(n), and their detailed description is included in equations (5.4). The decision variables describe the proportion of the section division between the final value of the range and the previous section point. They can assume values in the range from 0 to 1. Themaximum values of Pmax and the admissible Pnom result from the accepted assumptions P1 =x(1)(Pmax −Pnom) P2 =x(2)(Pmax −Pnom) P3 =x(3)(Pmax −P2) P4 =x(4)(Pmax −P3) (5.4) Two types of the membership functions have been used in the fuzzy controller, the triangular and trapezoid ones. The visualisation of the developed convention is pictured in Fig. 6. The presented solution has been used for description of the membership functions for both inputs so values of 8 variables have been looked for in the process of optimization. The ranges of the membership functions of the output signal (values of the set rotational speed of theZ1 engine) have been accepted as constants in accordance with Fig. 7. Both the unloading force andmovement of theZ1 drive should be taken into account during fuzzy controller optimization in the goal function. Due to the limit switches that are used in the device, themost important criterion connectedwith the position of theZ2 actuator is that none of the trolleys goes beyond theworking range. This criterion is included in the form of a penalty function. Moreover, it is important to ensure fluent regulation of the unloading force in order to ensure the biggest patient’s comfort. It is achieved by limiting the second derivative of this value in terms of time. Theminimization of such a component also affects the energy efficiency Computational optimization and implementation of control system... 1187 Fig. 6. Visualization of the developedmethod of describing the input membership functions Fig. 7. Output membership function for rotational speed of the servomotorZ1 of the device. The inclusion of all the aspects discussed above has allowed for formulating a multi-objective goal function with the internal penalty function FCfuzzy =(1−w1−w2) T ∫ 0 |∆UNL(t)| dt+w1 T ∫ 0 d2|∆UNL(t)| dt2 dt +w2 T ∫ 0 |ωZ1(t)| dt+ T ∫ 0 |∆UNL(t)|P(t) dt→min (5.5) where P(t)= { 0 if Z2disp < 40mm 104 if Z2disp > 40mm (5.6) The optimization of both controllers has been carried outwith the use of theOptimizationTool- boxmodule that is implemented in theMATLAB softwaremaking use of a hybrid optimization method combining a genetic algorithm with a gradient algorithm for continuous optimization with limitations (fmincon function). The best individual created in the genetic algorithm con- stitutes the starting point for calculations with the gradient method in the calculations carried out. The research carried out has shown that it is most advantageous to use the PD controller for steering theZ2 drive with the following settings: P =57,13 I =0 D=0.755 The case in which the weight factor w1 = 0.005, w2 = 0.02 has been considered the best one after analysis of the set of paretooptimal solutions obtained during the optimization of the fuzzy controller. The obtained distribution of the membership functions and a graphic interpretation of the output signal (as a surface) are shown in the three consecutive figures. 1188 G. Gembalczyk et al. Fig. 8. Input membership function for unloading the force error Fig. 9. Input membership function for displacement of theZ2 actuator Fig. 10. Surface of the fuzzy logic controller output signal 6. Results of experimental research In order to verify the developed algorithm controlling the drives of the unloading system, a series of experiments has been conducted. In the presented results, the operation of the BWS drive systems has been tested during walking on a treadmill with a constant speed 2km/h, but the set value of the unloading force has been changed and equaled to 50, 120, 180, 300 and 180N, respectively. The charts show the registered value of the unloading force (Fig. 11), rotational speed of the servomotorZ2 (Fig. 12), displacement of the actuator in theZ2 drive system (Fig. 13) and the rotational speed of the servomotor Z1 (Fig. 14). Computational optimization and implementation of control system... 1189 Fig. 11. Registered value of the unloading force as a function of time Fig. 12. Registered value of theZ2 servomotor rotational speed as a function of time Fig. 13. Registered linear displacement of theZ2 actuator as a function of time Fig. 14. Registered value of theZ1 servomotor rotational speed as a function of time 7. Conclusions Thepaperpresents amechatronic device supporting theprocess ofwalking re-education inwhich the body weight support system cooperates with the training treadmill, and has the possibility of making follow-up movements with side movements of the patient. The use of a fuzzy controller makes it possible to develop a universal steering system that ensures the maintenance of a constant unloading force value and synchronizes the operation of 1190 G. Gembalczyk et al. the winding drum with the SEA drive type independently of the exercise done. The proposed solution is particularly beneficial for physiotherapists supervising a patient because it requires introduction of the unloading force value only. It is a particularly important advantage as the simplicity of handling is expected. As the charts show, the drive of theZ1 winding drum is started only in the situation when the Z2 actuator is close to the limit switches; it is not, however, used during normal walking on the treadmill. It is an advantage in economic terms. Testing the device by a healthy person only is a certain limitation of the conducted research; however, it is a beneficial situation for the verification of the steering system operation as the exercising person can move in a more dynamic way. References 1. Cao J., Xie S.Q., Das R., Zhu G.L., 2014, Control strategies for effective robot assisted gait rehabilitation: the state of art and future prospects, Medical Engineering and Physics, 36, 12, 1555-1566 2. Chen G., Chan C.K., Guo Z., Yu H., 2013, A review of lower extremity assistive robo- tic exoskeletons in rehabilitation therapy, Critical Reviews in Biomedical Engineering, 41, 4-5, 343-363 3. Dragunas A.C., Gordon K.E., 2016, Body weight support impacts lateral stability during treadmill walking, Journal of Biomechanics, 49, 13, 2662-2668 4. Duda S., Gąsiorek D., Gembalczyk G., Kciuk S., Mężyk A., 2016,Mechatronic device for locomotor training,Acta Mechanica et Automatica, 10, 4, 310-315 5. Duda S., Gemalczyk G., 2016, Experimental study on the fuzzy-PID hybrid control algorithm for unloading system in mechatronic device for gait re-education, Proceedings of VII European Congress on Computational Methods in Applied Sciences and Engineering, 6567-6573 6. Duda S., Gembalczyk G., Switonski E., 2017, Design study and development of mechatro- nic treadmill for gait reeducation, Engineering Dynamics and Life Sciences. Proceedings of 14th International Conference Dynamical Systems Theory and Applications 2017, 183-191 7. Frey M., Colombo G., Vaglio M., Bucher R., Jörg M., Riener R., 2006, A novelmecha- tronic bodyweight support system,Neural Systems and Rehabilitation Engineering, 14, 3, 311-321 8. Gniłka J., Mężyk A., 2017, Experimental identification and selection of dynamic properties of a high-speed tracked vehicle suspension system, Eksploatacja i Niezawodność – Maintenance and Reliability, 19, 1, 108-113 9. Hidler J., BrennanD., Black I., NicholsD., BradyK., 2011, ZeroG:Overground gait and balance training system, Journal of Rehabilitation Research and Development, 48, 4, 287-298 10. Jurkojć J., Wodarski P., Bieniek A., Gzik M., Michnik R., 2017, Influence of changing frequency and various sceneries on stabilometric parameters and on the effect of adaptation in an immersive 3D virtual environment,Acta of Bioengineering and Biomechanics, 19, 3, 129-137 11. KoceskaN.,Koceski S., 2013,Review: robot devices for gait rehabilitation, International Jour- nal of Computer Applications, 62, 13, 1-8 12. Koenig A., Omlin X., Bergmann J., Zimmerli L., Bolliger M., Müller F., 2011, Con- trolling patient participation during robot assisted gait training, Journal of Neuroengineering and Rehabilitation, 8, 14-25 13. Kot A., Nawrocka A., 2012, Balance platform system dynamic properties, Journal of Vibroen- gineering, 14, 1, 178-182 14. Lünenburger L., Colombo G., Riener R., 2007, Biofeedback for robotic gait rehabilitation, Journal of Neuroengineering and Rehabilitation, 4, 1 Computational optimization and implementation of control system... 1191 15. Mehrholz J., Pohl M., Elsner B., 2014, Treadmill training and body weight support for walking after stroke, [In:]Cochrane Database of Systematic Reviews, 1, JohnWiley & Sons, Ltd 16. Mężyk A., Klein W., Fice M., Pawlak M., Basiura K., 2016,Mechatronic model of conti- nuous miner cutting drum driveline,Mechatronics, 37, 12-20 17. Miądlicki K., Pajor M., 2015, Real-time gesture control of a CNC machine tool with the use Microsoft Kinect sensor, International Journal of Scientific and Engineering Research, 6, 538-543 18. Mignardot J.B., Le Goff C.G., van Den Brand R., 2017, Amultidirectional gravity-assist algorithm that enhances locomotor control in patients with stroke or spinal cord injury, Science Translational Medicine, 9 19. PajorM., Herbin P., 2015, Exoskeleton of upper limb –model using realmovement parameters (in Polish),Modelowanie Inżynierskie, 26, 57, 40-46 20. Pratt G.A., Williamson M.M., 1995, Series elastic actuators, IEEE International Conference on Intelligent Robots and Systems, 399-406 21. Querry R.G., Pacheco F., Annaswamy T., Goetz L., Winchester P.K., Tansey K.E., 2008, Synchronous stimulation and monitoring of soleus H reflex during robotic body weight- supported ambulation in subjects with spinal cord injury, Journal of Rehabilitation Research and Development, 45, 1, 175-186 22. Raczka W., Sibielak M., Kowal J., Konieczny J., 2013, Application of an SMA spring for vibration screen control, Journal of Low Frequency Noise, Vibration and Active Control, 32, 1-2, 117-131 23. Reinkensmeyer D.J., Dietz V., 2016, Neurorehabilitation Technology, Springer, International Publishing 24. Riener R., Lünenburger L., Maier I.C., Colombo G., Dietz V., 2010, Locomotor training in subjects with sensori-motor deficits: an overview of the robotic gait orthosis lokomat, Journal of Healthcare Engineering, 1, 2, 197-216 25. Robinson D.W., Pratt J.E., Paluska D.J., Pratt G.A., 1999, Series elastic actuator deve- lopment for a biomimetic walking robot,Proceedings of IEEE/ASME International Conference on Advanced Intelligent Mechatronics, 561-568. 26. Sapiński B., Rosół M., Węgrzynowski M., 2016, Investigation of an energy harvesting MR damper in a vibration control system, Smart Materials and Structures, 25, 12, 125017, 1-15 27. Snamina J., Kowal J., Orkisz P., 2013,Active suspension based on lowdynamic stiffness,Acta Physica Polonica A, 123, 6, 1118-1122 28. Xu W.J., 2012, Permanent magnet synchronous motor with linear quadratic speed controller, Energy Procedia, 14, 364-369 29. ZhaoC.S., ZhuS.J.,HeQ.W., 2007,Fuzzy-PIDcontrolmethod for two-stagevibration isolation system, Journal of Theoretical and Applied Mechanics, 45, 1, 171-177 Manuscript received January 16, 2018; accepted for print June 11, 2018