International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:16 No: L. J. Wei, A. Z. Hj Shukor, M. H.

International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:16 No:01 54 Investigation on the Effects of Outer-Loop Gains, Inner-Loop Gains and Variation of Parameters on Bilateral Teleoperation Control System Using Geared DC-Motor L. J. Wei, A. Z. Hj Shukor, M. H. Jamaluddin Abstract This paper presents on the variation of parameters, bandwidth between the disturbance observer (DOB) and reaction torque observer/reaction force observer (RTOB)/(RFOB), and the gains at the outer-loop/performance controller effects to the bilateral teleoperation control system. These design constraints affect during the design of bilateral teleoperation control system. Somehow the stability, robustness and performance of the DOB and RTOB based robust motion control system influenced by these design constraints. Similarly, the position and force control system also interconnected with the DOB and RTOB based robust motion control system. It is crucial to acknowledge the design constraints in order to design a bilateral control system. Moreover, improving the degree of reproducibility and operationality are important in bilateral control system. It indicates the transparency of the bilateral teleoperation control system which can be considered for further development. Thus, the aim of this paper is to conduct experiments to analyze on the design constraints of the robust motion control system in bilateral teleoperation control system. From these experiment results, the stability, robustness and performance of the DOB and RTOB based robust motion control system are validated. Moreover, improvement in the degree of reproducibility and operationality of bilateral teleoperation control system by the design constraints are analyzed. Index Term disturbance observer (DOB), reaction torque observer (RTOB), parameter variation, position controller, force controller, stability, robustness, performance, bilateral teleoperation control system. I. INTRODUCTION ALTHOUGH the application of DOB based motion control system has been implemented for almost three decades, there are yet still more space to analyze and improve the design control methods [1]. During the designing the DOB based motion control, parameters variation, bandwidth between the DOB and RTOB, and the gains at the outer-loop/performance controller influence not just robustness, but stability and performance of the motion control system [1], [2]. The outerloop consists of performance controllers which are position controller and force controller Furthermore, inner-loop consists of disturbance observer (DOB) and reaction torque observer (RTOB). These design constraints affect during the design of motion control system. Unlike in this paper, the analysis on the design constraints are investigated onto the bilateral teleoperation control system. Generally, four channel bilateral controller is well known to achieve high transparency and stability in bilateral teleoperation. It consists hybrid of position and force in acceleration dimension based on disturbance observer. Thus, it is critically important that all four channels must properly use in achieving high performance and stability sense of accurate transmission of remote impedances to the human operator. This is where the reproducibility and operationality of the bilateral teleoperation control system step in [3]. Basically, the robustness of acceleration control on bilateral teleoperation is achieved by DOB [2], [4] [6]. Estimation of disturbance able to improve transparency and robustness of bilateral control. The bandwidth of DOB is set as high as possible to estimate and suppress the disturbances in a wide range of frequency. However, there are noise and robustness constraints in practical [1]. Another controller is designed at the outer-loop to obtain performance of the system. The robustness from inner-loop and performance from outer-loop are designed independently from each other. This control structure is called as two-degree of freedom control [4]. However, in practical, both outer and inner-loop are related in terms of robustness. Furthermore, (RFOB)/(RTOB) has been implemented to enable force feedback without force/torque sensor [7]. It is an application of DOB and used to estimate the environmental impedance. DOB can increase the bandwidth of reaction force information in wide bandwidth and the ability of the bilateral control is improved [8]. The RTOB is designed by subtracting the external disturbances and system uncertainties from input of a DOB. Thus the DOB and RFOB are almost similar however, only the latter has a model based control structure, means it requires the exact model of plant and external disturbance [2]. Nevertheless, good sense of touch must be realized in the bilateral teleoperation control system. Thus, during the design of a bilateral teleoperation control system reproducibility and operationality are taken into account. Reproducibility is the degree of reproduction of environmental impedance in master side which is the fundamental motive in bilateral teleoperation control system. Operationality is degree of operational force which human operator feels besides reaction force from the environment which desired for comfort operation for human operator. In the case of reproducibility, the position controller gain plays a big role in the sense of touch. The higher the gain, the better the degree of reproducibility [3]. However, of course the position controller gain cannot freely increase due to noise, sampling time, and so on. In paper [3] and [9], a further study of the force controller and variation of nominal inertia plays a vital role in operationality of bilateral teleoperation control system, respectively. It stated that high in force gain, or small in nominal inertia,

International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:16 No:01 55 improve the operationality. In addition, the parameter variations contact motion using the proposed experiments. Moreover, affect the stability of the system. Theoretically, the stability of Section VIII concludes the outcome of this paper and discussion the system is analyzed by root locus, and as the nominal inertia, on further work. increase, instability occurs in the system. In this II. SINGLE LINK PLANAR ROBOT reproducibility and operationality section, the validation of this MANIPULATOR BILATERAL research section is confirmed by experimental results. In brief, TELEOPERATION CONTROL SYSTEM these analyses are effective in improving the performance and The haptic sensation can be realized by implementing the stability of bilateral teleoperation control system. bilateral control technique. By using such method, the force In this paper, the single link rotatory planar bilateral sensation occurs in the slave environment can be perceive by teleoperation control system is used. Generally, there are the master s control side and it goes both ways. For the precise position control and force control system in a bilateral perception of force sensation, both torque and position teleoperation control system. Both control system is influenced displacement should be transferred bidirectional. The total by the outer-loop gains, inner-loop gains, and variation of block diagram of bilateral motion control is summarized in Fig. parameters. The outer-loop consists of performance controllers 2.1 [10]. TABLE I shows the list of parameters definitions of which are position controller, and force controller, As this paper. usual, position controller consists of proportional, and TABLE I derivative gain,. The force controller consists of. LIST OF PARAMETERS DEFINITION Furthermore, inner-loop consists of disturbance observer gain, Parameter Description and reaction torque observer gain,. Link 1 In brief, to understand more deeply about the bilateral Real inertia teleoperation control system as motion control system, this Nominal inertia Motor inertia paper presents on the effects of outer-loop gains, inner-loop Load inertia gains, and variation of parameters to the bilateral teleoperation Gear ratio control system. These design constraints effect on the Nominal torque constant reproducibility and operationality of the bilateral teleoperation Proportional gain control system. The position control system of bilateral Derivative gain Force gain teleoperation control system is depended on,,, Position controller and. Equally important that the force control system is Force controller depends on,, and. Natural angular frequency This paper focuses on investigating the effects of the Damping coefficient parameter variation, bandwidth of the DOB and RTOB, and the Cut-off frequency of disturbance observer gains at the outer-loop/performance controller to the geared Cut-off frequency of reaction torque DC-motor in the bilateral teleoperation control system since observer most of the researchers study the bilateral teleoperation control Torque Angle system by using linear motor and non-geared DC-motor. Angular velocity Somehow, these design constraints that influence the stability, Angular acceleration robustness, and performance of the DOB and RTOB based Reference value robust motion control system. Next, experiments are conducted Response value to analyze on the design constraints of the robust motion control Disturbance value system in bilateral teleoperation control system. From these External value Master system experimental results, the stability, robustness and performance Slave system of the DOB and RTOB based robust motion control system are Common mode validated. Moreover, improvement in the degree of Differential mode reproducibility and operationality of bilateral teleoperation Estimated value control system by the design constraints are analyzed too. This paper is organized as follows; Section II introduces the bilateral teleoperation control system. Section III explains about the disturbance observer (DOB) and reaction torque observer (RTOB). Section IV explains the reproducibility and operationality of a bilateral teleoperation control system. Next, Section V explains on robustness, stability, and performance of the bilateral teleoperation control system. Section VI shows hardware setup of proposed bilateral teleoperation control system single link planar robot manipulator. Section VII discusses experiments and the results of the free motion and

International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:16 No:01 56 In order to fulfill the bilateral control system requirement and to comply the concept in a haptic system, this bilateral control system is equipped with disturbance observer (DOB) [6] and the reaction torque observer (RTOB) [7]. The differential mode of the system is position controlled using DOB while the common mode of the system is force controlled using RTOB. This system can automatically calculate force disturbance and external force that present in both master and slave system. The sensorless type of motion control system provide robustness of the system. Thus, the operator at the master system can feel the real sensation of the environment at the slave system even though the operator is not at the environment area. Fig. 2.2 shows the block diagram of single link bilateral control based Fig. 2.1. Total block diagram of bilateral motion control by acceleration control on acceleration control. In order to transfer information in haptic communication, the realization of position tracking and the law of action and reaction between master system and slave system are important. The represents the torque applied by human operator to the master manipulator while the represents the torque applied by the environment at the slave manipulator. Equation (2.1) represents the summation of the action torque from human operator and reaction force from environment should be zero. On the contrary, (2.3) represents that the position error between master position and slave position should be coming to zero. (2.1) (2.2) (2.3) Moreover, the basic concept of bilateral motion control system on both master and slave system are required to comply in its total acceleration in differential mode, and total force in common mode, [10]. However, the common mode and the differential mode are independent to each other. In order to have interaction between this two modes, the second order Hadamard matrix, is applied as modal decomposition as shown in Equation (2.8). Equation (2.4) to (2.7) has the characteristic of the second order Hadamard matrix (quarry matrix). (2.4) (2.5) [ ] [ ] [ (2.6) (2.7) [ ] ] (2.8) Fig. 2.2. Block diagram of single link bilateral control based on acceleration control In the differential mode, the system is fitted with the position controller while in the common mode, it is fitted with the force controller. The function of the position control system is to allow the tracking position system track the desired trajectory in the critically damp response. The position coefficient and velocity coefficient are set based in the natural angular frequency and a damping coefficient of the control system as show in Equation (2.9) and Equation (2.10) [11]. The force controller system has to maintain the contact stability between force at end-effectors and the force at the contact object [12]. They can be defined as; (2.9) (2.10) (2.11) (2.12)

International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:16 No:01 57 Fig. 2.3. Fundamental inertia and gear ratio In terms of the concept of inerter, input torque from motor drives the load inertia, through gear ratio of in Figure 2.3. Thus the total nominal inertia, and torque output, is to be calculate as; ( ) ( ) (2.13) (2.14) III. DISTURBANCE OBSERVER (DOB) AND REACTION TORQUE OBSERVER (RTOB) The DOB is a robust control tool that able to estimate the external disturbance and system uncertainties. DOB is also designed to cancel the disturbance torque as quickly as possible which act as disturbance compensation in a motion control system. Robust motion control is attained by using the disturbance observer, the robust motion controller makes a motion system to be an acceleration control system [8]. That is the reason DOB is implemented in order to establish robust acceleration controller [4]. The output of DOB is the friction effect under the constant angular velocity motion in the mechanism. A robust system means that the system is insensitive to the external disturbance and parameter variations. It can obtain wider bandwidth than force sensor due to settling sampling time and observer gain by using DOB [7]. DOB is very effective for motion control and robust to both parameter uncertainties and unknown disturbance [13] and also provide a DOB based robust position control system[1]. The feedback of estimated disturbance in the inner-loop is to obtain the robustness of the motion control system [13]. On the contrary, the outer-loop is to estimate the external forces to realize force servoing. Moreover, the PD controller is designed in outer-loop so that the performance requirements of the motion control system are satisfied [13]. This kind of structure is a two degree of freedom control [4]. However, in practical, both outer and inner-loop are related in terms of robustness. Fig. 3.1 shows the block diagram of joint space based disturbance observer and reaction torque observer to compensate the disturbance effect within the motor plant and estimate the external torque from both the master and slave manipulators, respectively. Fig. 3.1. Block diagram of joint space based disturbance observer and reaction torque observer where; (3.1) (3.2) (3.3) (3.4) (3.5) Coulomb friction; Viscous friction; Self-inertia variation; Variation of torque coefficient; Load torque; In Equation (3.1), the first term and second term are the torque pulsations due to self-inertia variation and variation of the torque coefficient of the motor, respectively. The third term and the fourth term denoted the coulomb and the viscous friction, respectively. The last term is the reaction torque caused by external torque. The disturbance torque is estimated from the current reference and a velocity response. The estimated torque, is estimated using Equation (3.6) where; (3.6) (3.7) is the DOB low-pass filter (LPF) and cut-off frequency. is a

International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:16 No:01 58 ( ) (3.8) By using Equation (3.6), a realization of robust motion control is attained. The bandwidth of the DOB low-pass filter as in Equation (3.7) is set as high as possible to estimate a wide frequency range of disturbance. However, it is limited by practical and robustness constraints [1]. Furthermore, by subtracting the external disturbances and system uncertainties from input of a DOB, it can estimate the reaction torque applied to the system. It is necessary important to identify them as precisely as possible. This process is called as reaction torque observer (RTOB) [7] as shown is Fig. 3.1. Equation (3.9) shows that the reaction torque observer is estimated through first-order LPF. RTOB can estimates external torque without torque sensor because it has many drawbacks. The study of comparison between force sensor and reaction force observer based on force control system has been analyzed [14]. where; (3.9) (4.4.) Thus, there is a perfect reproduction of environment impedance in master side. However, is not satisfied. Thus, according to the Equation 4.4, can be close to zero by increasing or using a smaller. In paper [3] and [9], a further study of the force controller and variation of nominal inertia, respectively, play a vital role in operationality of bilateral teleoperation control system. It is stated that high value of or small value of nominal inertia improves the operationality. Nevertheless, the reproducibility and operationality can be improved simultaneously by lowering the nominal inertia less than real inertia. In addition, the parameter variations affect the stability of the system. Theoretically, the stability of the system is analyzed by root locus, and as the nominal inertia increases, instability occurs in the system. As the force gain, indicates the operationality of the system during free motion, position controller which consists of and indicate the reproducibility of the system during contact motion. Thus, a few experiments are conducted to investigate the effect of position gains, force gains, and nominal inertia on reproducibility and operationality of the bilateral teleoperation system. is the DOB low-pass filter (LPF) and cut-off frequency. IV. REPRODUCIBILITY AND OPERATIONALOTY is a (3.10) In [3], reproducibility and operationality are defined as and, respectively. Reproducibility is the degree of reproduction of environmental impedance in master side which is the fundamental motive in bilateral teleoperation control system. Operationality is degree of operational force which human operator feels besides reaction force from the environment which is desired for comfortable operation for a human operator. In an ideal state; (4.1) (4.2) which means there is a perfect reproduction of environment impedance in master side followed by zero operational force. However, this happens if the bandwidth of the DOB and RTOB is wide enough or infinity as the disturbance can be perfectly suppressed with the external force precisely estimated. With this condition, the and have become; (4.3) V. ROBUSTNESS, STABILITY AND PERFROMANCE In bilateral teleoperation control system, the robustness, stability, and performance must be considered. Each gains and parameters play an important role when designing a bilateral teleoperation control system. Thus in this paper, experiments are conducted to investigate the effect of gains and parameters to the bilateral teleoperation control system. The robustness of the position control system depends on inner-loop (DOB) whereas the performance depends on the outer-loop (position controller) by using acceleration based controller. However, as mentioned in [1], the robustness and performance of a DOB based motion control system are adjusted in the inner and outer loops independently, which is not true. The robustness of a position control system depends on the DOB as well as the outer loop performance controller. By increasing the outer-loop gain, the robustness of the position control system improved, however, the robustness of inner-loop becomes more sensitive to disturbance at high frequency such as noise. Besides, increasing the outer-loop gain can cause vibration due to attracting high frequency dynamics, saturation, and energy consumption. When a DOB is used, the system obtained a good robustness in a wide range of frequency, yet its performance is limited by the dynamic characteristics of the first order of DOB. Increasing the bandwidth of DOB improves the robust stability. DOB cannot estimate high frequency disturbance precisely, so the robustness of the system deteriorates. Thus the disturbance can be suppressed accurately if it stays within the bandwidth of

International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:16 No:01 59 VI. HARDWARE SETUP OF BILATERAL TELEOPERATION CONTROL SYSTEM PLANAR ROBOT MANIPULATOR DOB. Thus there is a trade-off between the robustness and noise response to determine the bandwidth of DOB [1]. The disturbance observer gain also does not make the system unstable. The bandwidth of DOB supposed to set as high as possible to compensate and estimate the disturbance in wide range of frequency. However, it is limited by practical and robust constraints[1]. These are dependent on sampling time and noise [3]. Furthermore, a DOB estimates external disturbances and system uncertainties in the inner-loop. The robustness of the force control system is achieved by feeding back the estimated disturbances. However, system uncertainties should be identified as a priori to design an RTOB in the outer-loop. Although the structures of a DOB and an RTOB are quite similar, only the latter is a model based control method, means it requires the exact model of plant and external disturbance which is the most challenging issue in its design [2]. In general, an RTOB is considered as a feed-forward control structure to simplify the analysis. However, it is not true. Nevertheless, the design parameters of a DOB and an RTOB change not only the performance, but also the stability of the force control system. The setting of different bandwidth of DOB and RTOB able to form a phase lead-lag compensator, and the stability and performance of force control system can be improve by increasing the bandwidth of RTOB [1]. Moreover, the stability of the force control system significantly changes due to the imperfect identification of inertia and torque coefficient. The torque coefficient, should be precisely identified in the design of RTOB to improve the performance. The performance of RTOB changes significantly by the imperfect identification of torque coefficient. However, the inertia identification can be neglected due to small acceleration in force control [2]. Thus, the stability of the force control system can improve the stability by designing. However, in practice, although torque coefficient identification can be achieved precisely, identification of inertia may not be a simple task, e.g., the inertia of a multi-body system is quite complex and non-linear. Therefore, not only the performance, but also the stability of the force control system deteriorates significantly by the imperfect identification of inertia and torque coefficient [1]. Thus, when it mentioned stability of force control, not just the nominal inertia, affect the stability, force gain,, and bandwidth of RTOB also affect the stability of the force control system. As increase, the stability of the force control system deteriorates. The design parameters of the DOB and RTOB drastically effect the stability of the force control system. Increase the bandwidth of RTOB increase the stability. Moreover, if the nominal inertia is lower than inertia, the stability improve in the design of RTOB [1]. In short, the stability of force control to estimate the reaction torque from the environment can be improved by designing the force control system using and. There are two sets of planar robot manipulators. Each joint is actuated by a planetary geared DC-Micromotor with incremental encoder. It has 5000ppr (pulse per revolution) before gearhead. The links are designed with a 0.12m each with a base attached to a platform to prevent any unwanted vibration. The link can be either used to operate the system by human operator on master side or to the environmental contact on the slave side. The links at both master and slave side are horizontal orientated that provided zero gravitational effects. Thus, only disturbance effect and frictional force presented in the gearhead and motor. Moreover, the main purpose of this setup is to investigate the operationality torque and reproducibility of this propose geared bilateral teleoperation control system. Fig. 6.1 shows the picture of the manipulator assembled while Fig. 6.2 shows the parts of a single manipulator. Fig. 6.1. Overall hardware manipulator setup Fig. 6.2. Parts of a single manipulator Planetary gearhead able to provide higher torque for a low torque DC-Micromotor. Moreover, the backlash is crucial to the haptic application where it can affect the performance of the bilateral system. However, the backlash of this gearhead is less

International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:16 No:01 60 than 1º, small enough to avoid degrading the performance of the bilateral system. The output of master and slave manipulator is position that measured by the encoders mounted at the back of each motor shaft. The velocity response is obtained by derivative of position response and filter away the noisy signal by LPF within the computer software. Within the software, the processed data is set to analog voltage reference signal to the motor driver. The reference value represents the desired current that the motor driver injected to the motor. The motor torque is directly Fig. 6.4. Micro-Box 2000 x86 Based Real-Time System proportional to the motor current [15]. Fig. 6.3 shows the overall diagram of the system. Fig. 6.5. Micro-Box 2000 I/O pins Fig. 6.3. Overall diagram of the system Micro-Box 2000 x86 Based Real-Time System is an affordable and robust platform for rapid control prototyping applications as shown in Fig. 6.4. It is rugged, high performance and can fulfil real-time analysis and control system testing needs. The control system for these experiments is designed using Simulink which is integrated to the Micro-Box and allow real-time modeling and simulation of control systems which is important to plot real time data. Moreover, the sampling time of this Micro-Box can go up to 1ms. Specifications of the Micro-Box: 1. Rugged, high performance industrial PC. Fan less, low-power consumption design (22W typical) Support for all standard PC peripherals, includes external floppy. Sturdy, compact size. 2. I/O-equipped with AD/DA, Encoder, CAN, and DI/O modules. 3. Onboard Celeron M 1GHz/256 MB DDR RAM, 64MB compact flash RAM (expandable to 1GB). 4. Stand-alone operation with xpc Target Embedded OptionTM. Users can write the Simulink model onto a CF card without an Internet connection. 5. I/O pins specifications in Fig. 6.5. VII. EXPERIMENTAL RESULTS AND DISCUSSION At each experiment, an action 1Nm torque input at each 1 second is applied at the master system. The reaction torque from the slave system is recorded. The position response and torque response from RTOB from each master and slave system are recorded and compared. During the experiments of free motion, the human operator operates the master handle while the slave handle is not constrained by environment or object. Next, during the contact motion, the slave handle is in contact with the environment or object. The experiments are conducted with the parameters shown in TABLE II. TABLE II PARAMETERS IN EXPERIMENT 7.1-7.4 Parameter Description Value Link 1 Nominal inertia Nominal torque constant Gear ratio A. Increment of Position Controller Gain In this section, fixed paramters for the experiments are shown in TABLE III while experiments are conducted with four different cases of position controller gains, and as shown in TABLE IV. This section, contact motion experiment are conducted according to the cases. Fig. 7.1 to 7.4 show the torque and position responses of both master and slave for each case. Fig. 7.5 shows the combination of all cases of torque responses at slave system. TABLE III PARAMETERS IN EXPERIMENT 7.1 Parameter Description Value Cut-off frequency of disturbance observer Cut-off frequency of reaction torque observer Force gain

International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:16 No:01 61 TABLE IV EXPERIMENTAL CASES IN EXPERIMENT 7.1 Case Case 1 Case 2 Case 3 Case 4 When the position controller gain increase, the position control system performance increased. This also led to increase in performance and stability in force tracking at force control system during contact motion. It is appropriate for good characteristic to set the gains as large as possible. However, as mentioned in Section 5.0, the gain cannot increase freely. These are dependent on sampling time and noise [3]. As mentioned above, the reproducibility is the fundamental motive in bilateral teleoperation which means the good bilateral system has a good reproduction of environment impedance at master side. When position controller gain is set to large value, reproducibility is remarkably progressed [3]. In contrast, the reproducibility deteriorated as the position gain small and higher position error occurred in contact motion as shown in Fig. 7.6. This position error between master and slave can lead to human operator cannot feel the sense of touch right. Thus, during the contact motion, the position and force are almost perfectly track when the position controller gain is large and the reproducibility is high. Thus, the human operator can feel sharp touch sense of hard object. However, the larger the environmental stiffness, the system becomes unstable [3]. Fig. 7.2. Case 2 (Contact motion) Fig. 7.3. Case 3 (Contact motion) Fig. 7.1. Case 1 (Contact motion)

International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:16 No:01 62 for each case. Fig. 7.12 show the combination of all cases of torque responses at slave system. TABLE V PARAMETERS IN EXPERIMENT 7.2 Parameter Description Value Proportional gain Derivative gain Cut-off frequency of disturbance observer Cut-off frequency of reaction torque observer TABLE VI EXPERIMENTAL CASES IN EXPERIMENT 7.2 Case Case 1 Case 2 Case 3 Case 4 Case 5 Fig. 7.4. Case 4 (Contact motion) When the force controller gain increase during the contact motion, the performance and stability of position and force tracking improved. The suppression of external disturbance also improved. However, as the force controller gain further increase, the performance and stability of position tracking deteriorates during the contact motion, respectively. The stability of the force control system also deteriorates as increases [2]. Thus, the force control gain cannot be increased freely due to the stability constraints. The responds of position and force control system become underdamped. It is appropriate for good characteristic to set the gains as large as possible. Fig. 7.5. Torque response at slave system Fig. 7.6. Position error (Contact motion) B. Increment of Force Controller Gain In this section, fixed paramters for the experiments are shown in TABLE V while experiments are conducted with five different cases of force controller gains, as shown in TABLE VI. This section, contact and free motion experiments are conducted according to the cases. Then, free motion experiment was conducted according to Case 1 and Case 4. Fig. 7.7 to 7.12 show the torque and position responses of both master and slave Fig. 7.7. Case 1 (Contact motion)

International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:16 No:01 63 Fig. 7.8. Case 2 (Contact motion) Fig. 7.10. Case 4 (Contact motion) Fig. 7.9. Case 3 (Contact motion) Fig. 7.11. Case 5 (Contact motion) Fig. 7.12. Torque response at slave system

International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:16 No:01 64 On the contrary, Fig. 7.13 to 7.14 show the torque and position responses of both master and slave for Case 1 and Case 4 during free motion, respectively. Fig. 7.14, the human operator has easily freely move the master system many motion in 10s as compare to Case 1 as shown in Fig. 7.13. Moreover, the operator felt there is no need extra operational force needed to apply during free motion as compare to Case 1. The amplitude of the operational force in Case 1 and Case 4 are about 0.3 Nm and 0.1 Nm, respectively. The operational force gains in Case 1 and Case 4 are 15 and 5 gains, respectively. Thus, these show that operationality of bilateral teleoperation system can be improved based on design so that operational force can be lessen. Position response of master and slave of all cases are well tracked. As mentioned above, operationality is desired for a comfortable operation which means the human operator able to control the bilateral system without or with small operational force. Operational force means the force the human operator feels in addition to real environment force. When force controller gain is large, the operationality is remarkably progressed [3]. In free motion, as the force controller gain increase, the operational force is reduced. Means the human operator able to handle the master system to control slave system without applying any extra force during free motion. Moreover, it is possible to obtain stable response by considering as PD controller using derivative information of force. The extra force is called as operational force. However, there is unignorably operational force appeared which produced by friction of the motor. However, the gain cannot increase freely. These are dependent on sampling time and noise [3]. Fig. 7.14. Case 4 (Free motion) C. Increment Bandwidth of and In this section, fixed paramters for the experiments are shown in TABLE VII while experiments are conducted with four different cases of bandwidth of DOB and RTOB gains, and as shown in TABLE VIII. This section, contact motion experiment is conducted according to the cases. Fig. 7.15 to 7.18 show the torque and position responses of both master and slave for each case. Fig. 7.19 show the combination of all cases of torque responses at slave system. TABLE VII PARAMETERS IN EXPERIMENT 7.3 Parameter Description Value Proportional gain Derivative gain Force gain TABLE VIII EXPERIMENTAL CASES IN EXPERIMENT 7.3 Case Case 1 Case 2 Case 3 Case 4 Fig. 7.13. Case 1 (Free motion) As shown is Fig. 7.15 to 7.18, as the and increase together at same rate, the stability and performance of position and force tracking is improved. When the bandwidth of both DOB and RTOB is narrow, the responses of the position tracking is slow. Moreover, force responses have a sluggish response. However, when the bandwidth gains of and increase from 300 rad/s and above, the stability of force tracking deteriorates as shown in Fig. 7.18. The force control starts to overshoot and oscillate. This is because as the bandwidth of DOB increases, the robustness improves but the stability deteriorates and vice versa. In this paper, the

International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:16 No:01 65 environment stiffness is high, thus the problem is not severe by increasing the bandwidth of DOB, but the stability may deteriorate when environment object is soft. However, by decreasing the robustness of the system, means decrease the bandwidth of DOB, the stability improve and the stability improvement become more dominant when environmental stiffness is low. Thus the robustness of force control system should be design according to the stability constraint [2]. Generally to solve the instability in force control, the velocity feedback gain is enlarged, however the manipulator s response becomes slow and vice versa [8]. This paper [3] mentioned the system keeps stable in wideband sensing. Means, the wider the bandwidth of RTOB, the stable the system becomes. It is also proved that bandwidth of force sensing of the RTOB is wider than the force sensor which is necessary for stable force control [1], [8]. Moreover, the performance and stability of the force control also improved as increasing in bandwidth of RTOB. Fig. 7.16. Case 2 (Contact motion) Fig. 7.15. Case 1 (Contact motion) Fig. 7.17. Case 3 (Contact motion)

International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:16 No:01 66 When and are set at the different value, a phase lead-lag compensator is formed which can be improve the stability and performance of the robust force control, by increasing bandwidth of RTOB, [1]. If: When : no phase compensator is obtained. 1. The stability of the robust force control system deteriorates as the environmental stiffness/damping or, increases/decrease. This phenomenon is tested in [5], [1]. When : a phase lag compensator is obtained. 1. The stability and performance of the robust force control system degraded. 2. The force control system and the position control system have overshoot when contacting to hard environment. Fig. 7.18. Case 4 (Contact motion) When : a phase lead compensator is obtained. 1. The stability and performance of the robust force control system improved. 2. The force control system is more merely the same when. However, the position and force tracking are better. Thus, the phase lead compensation based on disturbance observer makes the stability of the force control system improve by designing [2]. Fig. 7.19. Torque response at slave system D. When In this section, fixed paramters for the experiments are shown in TABLE IX while experiments are conducted with three cases of bandwidth of DOB and RTOB as shown in TABLE X. This section, contact motion experiment is conducted according to the cases. Fig. 7.20 shows torque response at slave system while Fig. 7.21 to 7.22 show the position responses of master and slave system for each case, respectively. Fig. 7.20. Torque response at slave system TABLE IX PARAMETERS IN EXPERIMENT 7.4 Parameter Description Value Proportional gain Derivative gain Force gain TABLE X EXPERIMENTAL CASES IN EXPERIMENT 7.4 Case Case 1 Case 2 Case 3 Fig. 7.21. Position response at master system

International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:16 No:01 67 E. Variation of Fig. 7.22. Position response at slave system In this section, fixed paramters for the experiments are shown in TABLE XI while experiments are conducted with three different cases of nominal inertia as shown in TABLE XI. This section, contact and free motion experiments are conducted according to the cases. TABLE XI PARAMETERS IN EXPERIMENT 7.5 Parameter Description Value Proportional gain Derivative gain Force gain Cut-off frequency of disturbance observer Cut-off frequency of reaction torque observer Fig. 7.23. Case 1 (Free motion) TABLE XII EXPERIMENTAL CASES IN EXPERIMENT 7.5 Case Case 1 Case 2 Case 3 Fig 7.23 to 7.25 show the torque and position responses of both master and slave for each case during free motion. In Case 1, the human operator easily moves the master system freely for many motion in 10s as compare to Case 2 and Case 3. Moreover, the operator felt there is no need extra operational force needed to apply during free motion as compare to Case 2 and Case 3. The amplitude of the operational force in Case 1, 2, and 3 are about 0.06 Nm, 0.2 Nm, and 0.25 Nm, respectively. The operational force gains in Case 1, Case 2, and Case 3 are 3, 10, and 12.5, respectively. As the nominal inertia, increase, the operational force to operate the master system increase. Thus, these show that operationality of bilateral teleoperation system can be improved based on smaller nominal inertia design so that operational force can be lessen. Position response of master and slave of all cases are well tracked. Fig. 7.24. Case 2 (Free motion)

International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:16 No:01 68 Fig. 7.27. Torque response at slave system Fig. 7.25. Case 3 (Free motion) Fig. 7.28. Position error (Contact motion) Fig. 7.26. Position error (Free motion) On the contrary, Fig. 7.26 shows the position error of between master and slave system during free motion of each case. Whereas, Fig. 7.27 and Fig. 7.28 show the torque response of slave system and position error of between master and slave system during contact motion of each case. In terms of stability of the force control system, it also affected by the design of nominal parameter. Thus, the stability and performance of force control to estimate the reaction torque from the environment can be improved by designing the force control system using, in the condition of as shown in Fig. 7.27 [2]. However, in term of stability of the position control system, it can be improved by design of, in the condition of. This can be proved in Fig. 7.26 and 7.28 as position error between master and slave system of free motion and contact motion decrease as increases. However, cannot increase freely due to the practical robustness constraints [1]. Overall, the variation of nominal inertia is design according to requirement which either to robust position control or robust force control. VIII. CONCLUSION When the position controller gain increase, the position control system performance increased. This also led to increase in performance and stability in force tracking at force control system during contact motion. Furthermore, during the contact motion, the position and force are almost perfectly track when the position controller gain is large and the reproducibility high. Then again, the gain cannot increase freely. These are dependent on sampling time and noise When the force controller gain increase during the contact motion, the performance and stability of position and force tracking improved. The suppression of external disturbance also improved. However, as the force controller gain further increase, the performance and stability of position tracking deteriorates during the contact motion, respectively. The stability of the force control system also deteriorates as increases. Thus, the force control gain cannot be increased freely due to the stability constraints. Besides, when force controller gain is large, the operationality is remarkably progressed [3]. In free motion, as the force controller gain increase, the operational force is reduced. However, the gain cannot increase freely. These are dependent on sampling time and noise [3]. On the contrary, bandwidth of DOB increases, the robustness improved but the stability deteriorates and vice versa and the performance of the force control also improved as increasing in bandwidth of RTOB. As the nominal inertia, increase, the operational force to operate the master system increase. Thus, these show that operationality of bilateral teleoperation system can be improved based on nominal inertia design so that operational force can be

International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:16 No:01 69 lessen. Likewise, the stability of force control to estimate the Force Cooperative Control of Multimanipulator Based on Workspace Disturbance Observer, Proc. IEEE 22nd Int. Conf. Ind. reaction torque from the environment can be improved by Electron. Control. Instrum., pp. 1873 1878, 1996. designing the force control system using and [13] O. Ozen, E. Sariyildiz, H. Yu, K. Ogawa, K. Ohnishi, and A.. Sabanovic, Practical PID Controller Tuning for Motion Control, As compare the robustness of position control system 2015 IEEE Int. Conf. Mechatronics, pp. 240 245, 2015. [14] E. Sarlyildlz and K. Ohnishi, A Comparison Study for Force depends not only on the DOB but also on the outer-loop Sensor and Reaction Force Observer based Robust Force Control performance controller. As the PD gain increase, the robustness Systems, 2014 IEEE 23rd Int. Symp. Ind. Electron., pp. 1156 of position control system improves same as when bandwidth of 1611, 2014. DOB increase. However, the gain cannot increase freely. These [15] A. Hace and K. Jezernik, Bilateral Teleoperation by Sliding Mode Control and Reaction Force Observer, Proc. IEEE Int. Symp. Ind. are dependent on sampling time and noise. The experimental Electron, pp. 1809 1816, 2010. results clearly show the validity of the proposals. Thus in the future, during the modeling of bilateral teleoperation control system using DOB and RTOB based motion control system, the robustness, stability and performance must be taken care. Those design constraints that discussed must be considered during the modeling and overcome the major problems of bilateral teleoperation control system. Furthermore, reproducibility and operationality are very important in order to increase the transparency of the bilateral teleoperation control system. These must be considering in designing bilateral teleoperation control system. ACKNOWLEDGMENT The Authors wish to express their thanks to the UTeM ZAMALAH scheme and UTeM PJP Grant (PJP/2015/FKE(1D)/S01392). REFERENCES [1] E. Sariyildiz and K. Ohnishi, Stability and Robustness of Disturbance-Observer-Based Motion Control Systems, IEEE Trans. Ind. Electron., vol. 62, no. 1, pp. 414 422, 2015. [2] E. Sariyildiz, S. Member, and K. Ohnishi, On the Explicit Robust Force Control via Disturbance Observer, IEEE Trans. Ind. Electron., vol. 62, no. 3, pp. 1581 1589, 2015. [3] W. Iida and K. Ohnishi, Reproducibility and Operationality in Bilateral Teleoperation, Proc. 8th IEEE Int. Work. AMC, pp. 217 222, 2004. [4] K. Ohnishi, M. Shibata, and T. Murakami, Motion Control for Advanced Mechatronics, IEEE/ASME Trans. Mechatronics, vol. 1, no. 1, pp. 56 67, 1996. [5] Y. Matsumoto, S. Katsura, and K. Ohnishi, An Analysis and Design of Bilateral Control Based on Disturbance Observer, pp. 802 807, 2003. [6] K. Ohnishi, N. Matsui, and Y. Hori, Estimation, Identification, and Sensorless Control in Motion Control System, Proc. IEEE, vol. 82, no. 8, pp. 1253 1265, 1994. [7] T. Murakami and K. Ohnishi, Torque Sensorless Control in Multidegree-of-Freedom Manipulator, IEEE Trans. Ind. Electron., vol. 40, no. 2, pp. 259 265, 1993. [8] S. Katsura, Y. Matsumoto, and K. Ohnishi, Modeling of Force Sensing and Validation of Disturbance Observer for Force Control, Proc. 29th Annu. Conf. IEEE Ind. Electron. Soc. IECON 03, vol. 54, no. 1, pp. 530 538, 2007. [9] K. Nishimura, T. Shimono, and K. Ohnishi, Improvement of Reproducibility and Operationality for Bilateral Control by Norninal Mass Design, pp. 1326 1331, 2006. [10] K. Ohnishi, S. Katsura, and T. Shimono, Motion Control for Real World Haptics, IEEE Ind. Electron. Mag., vol. 4, pp. 16 19, 2010. [11] M. H. Jamaluddin, T. Shimono, and N. Motoi, An Integration Method between Vision-based Disturbance Observer and Bilateral Haptic System for Robust Tracking of Target Object, IEEE 13th Int. Work. Adv. Motion Control (AMC), 2014, vol. 0, no. 1, pp. 723 728, 2014. [12] E. Leksono, T. Murakami, and K. Ohnishi, On Hybrid Position /