Received January 27, 2018, accepted February 19, 2018, date of publication February 27, 2018, date of current version April 4, 2018. Digital Object Identifier 10.1109/ACCESS.2018.2809860
Modular Development of Master-Slave Asymmetric Teleoperation Systems With a Novel Workspace Mapping Algorithm ZHENG CHEN 1,2 , (Senior Member, IEEE), SHUIFENG YAN3 , MINGXING YUAN BIN YAO1,4 , (Senior Member, IEEE), AND JINFEI HU1
1,
1 State
Key Laboratory of Fluid Power and Mechatronic Systems, Zhejiang University, Hangzhou 310027, China College, Zhejiang University, Hangzhou 310027, China 3 Huawei Enterprises Telecommunication Technologies Co., Ltd, Hangzhou 310052, China 4 School of Mechanical Engineering, Purdue University, West Lafayette, IN 47907, USA 2 Ocean
Corresponding author: Mingxing Yuan (
[email protected]) This work was supported in part by the National Natural Science Foundation of China under Grant 61603332 and Grant U1609211, in part by the Science Fund for Creative Research Groups of National Natural Science Foundation of China under Grant 51521064, and in part by the Youth Funds of the State Key Laboratory of Fluid Power and Mechatronic Systems, Zhejiang University.
ABSTRACT This paper proposes a modular teleoperation software platform based on the robot operating system (ROS), which is flexible and portable for different master and slave devices. The main issues for the modularity are to develop its control modules in MATLAB/Simulink and access ROS by the robotics system toolbox, which will facilitate the design, analysis, and adjustment of the control algorithm in teleoperation systems greatly. In addition, a case study for this kind of modular teleoperation system is carried out, which is composed of a Geomagic Touch as the master device and a robotic arm of the Rethink Baxter robot as the slave device. Due to the asymmetric structure property between the master and slave devices, the proper workspace mapping which can cover the whole workspace of the slave robot and simultaneously achieve an excellent mapping accuracy becomes a challenging issue. Different from the traditional methods, a hybrid workspace mapping algorithm is proposed, which uses the joint space mapping to cover the whole workspace of the slave device and the operating space mapping to conduct the elaborate manipulation. A smoothing switch law between the joint space mapping and the operating space mapping is also designed. Comparative experiments are carried out, and the results verify the effectiveness and excellent performance of the proposed design methodology. INDEX TERMS Teleoperation, workspace mapping, robot operating system (ROS), MATLAB/Simulink. I. INTRODUCTION
With the development of the modern industry, the requirements for the exploration work and operating environment where people are hardly to access directly are increasing, such as dealing with nuclear and toxic chemical materials in the nuclear and chemical industries, space and underwater exploration tasks and medical applications [1]–[4]. Although the robot technology has been developed greatly in the previous decades, it is still a challenging issue for an autonomous robot to perform these exploration and operation tasks intelligently and efficiently. To this end, the master-slave teleoperation system, which works in the human-machine interaction mode, becomes an effective solution. A typical teleoperation system architecture, as shown in Fig. 1, consists of a human operator, a master robot, the master controller, the 15356
FIGURE 1. Typical teleoperation structure.
communication channel, the slave controller, a slave robot and the environment [5]. In this framework, the human operator operates the master robot in the local site and generates
2169-3536 2018 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
VOLUME 6, 2018
Z. Chen et al.: Modular Development of Master-Slave Asymmetric Teleoperation Systems With a Novel Workspace Mapping Algorithm
the command trajectory which is then transferred to the slave robot site through the communication channel. In the remote site, the motion controller of the slave robot tracks the command trajectory from the the master robot site. Furthermore, the master robot is typically equipped with a haptic device which allows the human operator to perceive the tracking performance or the interaction with the environment in the slave robot site. Currently, many research projects that investigate developments of teleoperation systems with force feedback by employing haptic devices have been reported [3], [6]–[9]. However, most of them are set up for particular master and slave devices, and it is time-consuming to develop teleoperation systems with alternative master or slave devices in the absence of a modular software platform. To address this issue, Robot Operating System (ROS) [10], which is an open source software platform to provide the supports of building teleoperation systems, is developed in the robotics community recently. In this way, the data acquisition, signal transmission and hardware drivers are realized through ROS, and the controllers are developed by the C++ or Python language on the basis of ROS. However, it is well known that many control algorithms are implemented and tested on MATLAB/SIMULINK currently, and most of the control researchers and engineers are not much familiar with the programming skills using C++ or Python on the platform of ROS directly, which actually are not of much concern for the control researchers and engineers. In fact, those issues like data acquisitions and hardware drivers can be treated as low-level implementation problems. From the view of the control researchers and engineers, what they are concerned of is how to implement the control algorithms quickly and efficiently on a specific platform in the high level, for example, MATLAB/SIMULINK, but not the low-level issues above. In MATLAB/SIMULINK, plenty of control-related modules can definitely accelerate the control algorithms developments. As a result, MATLAB/SIMULINK has been widely adopted as the implementation or test platform of control algorithms in the control areas. As such, interfaces between MATLAB/SIMULINK and ROS, which allow the control researchers and engineers to design and implement the controllers on MATLAB/SIMULINK directly, are essential and will definitely facilitate the control systems developments greatly. On the other hand, for the master-slave asymmetric teleoperation systems, another challenging issue is to design a workspace mapping method to cover the whole workspace of the slave robot and perform elaborate operation tasks simultaneously. Ju et al. [8] use an Omni haptic device to operate an arm of the Rethink Baxter robot by a workspace mapping method with a constant scaling factor. However, to cover the workspace of the slave robot as much as possible, a large constant scaling factor is required in their method, which however cannot achieve an elaborate task operation. Salcudean et al. [11] propose a boundary drift mapping approach, which divides the workspace of the master into VOLUME 6, 2018
two parts. In one part, the constant scaling factor mapping is utilized to perform elaborate operation tasks, while in the other part, the velocity mapping is applied to cover the whole workspace of the slave robot. However, unexpected wrong switches between the two parts may exist in the actual operations. Mamdouh and Ramadan [12] propose a positionvelocity switch mapping scheme by using additional switch variables in the master device. In their approach, the additional switches are employed to avoid the misoperations. However, it is not intuitive and convenient for the operator to control the slave robot by using the velocity mapping. In this paper, a modular teleoperation software platform is proposed on the basis of Robot Operating System (ROS). In this platform, several interfaces between the MATLAB/SIMULINK and ROS are devised based on the robotics toolbox in MATLAB/SIMULINK. As such, control researchers and engineers can implement control algorithms on MATLAB/SIMULINK directly. The developed platform consists of four subsystems: the master module, communication module, control module built in MATLAB/SIMULINK and the slave module. Each module has public interfaces which use ROS topics to communicate with each other. For alternative master or slave devices, only minor changes like replacing the corresponding device drivers in the master or slave module are required. Additionally, all the control algorithms are implemented on MATLAB/SIMULINK, which facilitates the design, analysis and adjustment of teleoperation controllers greatly. Furthermore, a case study with this kind of modular teleoperation system which is composed of a Geomagic Touch as the master device and a robotic arm of the Rethink Baxter robot as the slave device is presented in this paper. In addition, a novel hybrid workspace mapping method is proposed to deal with the challenging issues existing in the master-slave asymmetric teleoperation systems. In the proposed workspace mapping method, the joint space mapping is employed to cover the whole workspace of the slave robot, while the operating space mapping is applied to perform the elaborate operation tasks. A transition algorithm is also designed to guarantee smoothing switches between the two mappings. Thus, the proposed workspace mapping method can cover the whole workspace of the slave robot and achieve the elaborate task operations simultaneously. Several experiments are carried out, and the experimental results validate the excellent performance of the proposed design paradigm. II. MODULAR DEVELOPMENT OF THE TELEOPERATION ROBOT PLATFORM
The developed modular teleoperation robot system, as shown in Fig. 2, integrates multiple hardware and software modules. A. HARDWARE PLATFORM
The Geomagic Touch device, which has 6 degree of freedoms (DOFs), is specified as the master device. Furthermore, a stylus with two buttons is equipped on the end-effector of the master device. One of the buttons is customized to 15357
Z. Chen et al.: Modular Development of Master-Slave Asymmetric Teleoperation Systems With a Novel Workspace Mapping Algorithm
FIGURE 3. The teltouch module structure.
FIGURE 2. The hardware and software structure of the teleoperation system.
used to render feedback force to the operator. If an alternative device is used as the master robot, only minor changes of the Touch device driver in the red box in Fig. 3 are required.
control the grippers of the slave robot while the other one is treated as the trigger in our workspace mapping method. One arm of the Rethink Baxter robot with 7 DOFs is set up as the slave robot. In addition, three computers and one router are also used to build a small local communication network. The Touch device and the Baxter robot are connected to the master computer and the router through the local area network (LAN), respectively. FIGURE 4. The communication module structure.
B. SOFTWARE PLATFORM
In the proposed modular teleoperation robot system, the software platform is built by integrating ROS and MATLAB/SIMULINK seamlessly. Specifically, ROS is a flexible platform for developing robot software. It is a collection of tools and libraries that aim to simplify the task of creating complicated robot behaviors across a wide variety of robotic platforms [10], [13]. The built-in and well-tested message transfer system of ROS enables each distributed module to communicate with each other by ROS topics. The proposed teleoperation system consists of four modules: the teltouch module running on the master computer, MATLAB/SIMULINK running on the control computer, the communication module running on the slave computer and the Baxter module running on the computer equipped on the Baxter robot. Each module exchanges information with each other via ROS topics.
2) COMMUNICATION MODULE
The communication module is the medium of the information exchange between the ROS system and MATLAB/SIMULINK. However, several data format conversions are necessary to guarantee the communication process between the ROS system and MATLAB/SIMULINK since the robotics toolbox in MATLAB/SIMULINK does not support data type like string which is however widely used in the message types of ROS. To this end, the communication module is developed to deal with this matter and its structure is shown in Fig. 4. In public interfaces, the communication module subscribes three topics to receive messages from MALAB/SIMULINK. Afterwards, some conversions such as adding some strings are operated on these messages. Finally, the processed messages are published to the three topics that can be recognized in the ROS system.
1) TELTOUCH MODULE
The teltouch module interacts with the Geomagic Touch device via the OpenHaptics API which is provided by the manufacturer of the Geomagic Touch device. In this way, the state information of the Geomagic Touch device can be acquired and the human operator is able to perceive the feedback force of the Touch device. The structure of the teltouch module is shown in Fig. 3. In the public interfaces, the teltouch module publishes the Touch device state such as joint angles, velocities and so on to the specific topic and also subscribes other two topics to receive messages that are 15358
3) MATLAB/SIMULINK MODULE
The MATLAB/SIMULINK module shown in Fig. 5 uses the robotics toolbox in MATLAB/SIMULINK to connect the blocks in MATLAB/SIMULINK to the ROS system. As such, The developed control modules can be run in MATLAB/SIMULINK directly, which facilitates the design, analysis and adjustment of the control algorithm greatly. With this modular design, features like messages, publishers and subscribers in ROS are masked as simple blocks in MATLAB/SIMULINK, which definitely improves the efficiency VOLUME 6, 2018
Z. Chen et al.: Modular Development of Master-Slave Asymmetric Teleoperation Systems With a Novel Workspace Mapping Algorithm
FIGURE 5. MATLAB/SIMULINK module structure.
of control algorithms developments and benefits the control system designers greatly. Finally, in order to build these public interfaces, the topic function of ROS is used. For example: ros::Publisher touchstate_pub =n.advertise ("/robot/teltouch/left/touchstate", 1) builds a message publisher of the specific topic ("/robot/teltouch/left/touchstate"), which is used in the public interfaces of the teltouch module. Additionally, a statement like ros::Subscriber torque_sub = n.subscribe (‘‘/touch/left/matlabtorque’’, 10, matlabtorque Callback) builds a message receiver of the specific topic ("/touch/left/matlabtorque"), which is also used in the public interfaces of the teltouch module. Similarly, other module public interfaces are designed according to this rule.
workspace mapping scheme is proposed to deal with this problem. The key idea of the proposed workspace mapping approach can be summarized as follows. First, the joint space mapping method is employed to cover the whole workspace of the slave robot and is applied first in the rough positioning phase. Next, after the rough positioning process is completed, the joint space mapping method is switched to the operating space mapping approach to perform elaborate manipulations. Typically, This hybrid workspace mapping scheme consists of three parts: 1) the joint space mapping method; 2) the operating space mapping method; 3) the transition algorithm between these two methods.
FIGURE 6. The baxter module structure.
4) Baxter MODULE
The Baxter module, whose structure is shown in Fig. 6, interacts with the actual Baxter robot to acquire the state information of the robot. As shown in Fig. 6, in the public interfaces, it receives messages from two topics that are used to describe the states of the robot’s joints and gripper. In addition, the Baxter module also publishes the joints and gripper states of the Baxter robot to two topics. Again, If an alternative robot is used as the slave device, only minor changes of the slave device driver in the red box in Fig. 6 are required. III. WORKSPACE MAPPING
In most of the teleoperation robot systems, structure discrepancies between the master and slave robot exist, which further cause the workspace differences between the master and slave robot. To this end, the workspace mapping from the master to the slave site is essential and must be devised in an intuitive, safe and efficient manner. In this section, a novel hybrid VOLUME 6, 2018
FIGURE 7. Joint space mapping relationship.
A. JOINT SPACE MAPPING METHOD
The joint space mapping method simply maps the joint position of the master device to that of the slave device. Specifically, as shown in Fig. 7, the third joint E0 of the Baxter robot is fixed to be zero since the Touch device has only 6 revolute joints while the baxter robot is equipped with 7 revolute joints. The remaining joints are mapped as the relationship listed in Fig. 7. Furthermore, since the rotating direction of each pair of the mapping joint between the master and slave robot is converse, an inverted operation on each mapping joint of the Touch device is required. In addition, as shown in Table 1, the joint rotation ranges of the Touch device differ from those of the Baxter robot joints. To this end, the mapping 15359
Z. Chen et al.: Modular Development of Master-Slave Asymmetric Teleoperation Systems With a Novel Workspace Mapping Algorithm
TABLE 1. The joint rotation range of the master and slave device.
The robotics toolbox in MATLAB/SIMULINK [14] is then employed to develop the forward kinematic models of the master and slave devices based on the DH parameters shown in Table 2 and Table 3. Furthermore, a closed-loop inverse kinematics algorithm [15] is applied to compute the inverse kinematics of the baxter robot. Specifically, the closed-loop inverse kinematics algorithm is represented as q˙ db = γ · JT (qdb ) · er ,
method is then formulated as θbMax − θbMin θb = θbMax − × (θt − θtMin ) θtMax − θtMin
(1)
where θb and θt are the joint angles of the Baxter robot and Touch device, respectively, while θbMax and θbMin are the maximum and minimum joint angles of the Baxter robot, respectively. Similarly, θtMax and θtMin represent the maximum and minimum joint angles of the Touch device, respectively. Taking into account the specific values in Table 1, Eqn. (1) becomes θbs0 θbs1 θbe0 θbe1 θbw0 θbw1 θbw2
= = = = = = =
1.7016 − 1.75 (θt1 + 0.97) 1.047 − 1.86 (θt2 − 0.05) 0 2.618 − 1.33 (θt3 + 2.4) 3.059 − 1.30 (θt4 + 2.7) 2.094 − 1.83 (θt5 + 1) 3.059 − 1.25 (θt6 + 2.4) .
(3)
where q˙ db represents the joint velocity solution of the inverse kinematics, and the joint position solution of the inverse kinematics qdb can thus be obtained by integrating q˙ db . er = xrb −xdb , in which xrb is the desired endpoint pose of the Baxter robot, while xdb represents the computed endpoint pose of the Baxter robot based on the joint position solution of the inverse kinematics. Practically, xdb can be obtained by substituting qdb into the forward kinematics of the Baxter robot. γ is a constant, and JT (qdb ) represents the transpose of the slave robot’s Jacobian matrix. Details of this algorithm are shown in [14]. C. TRANSITION ALGORITHM BETWEEN THESE TWO METHODS
(2)
TABLE 2. The DH parameters of the Touch device.
In this part, a transition algorithm to achieve smoothing switches between the two workspace mapping methods is developed and detailed as follows. 1) The joint space mapping method is switched to the operating space method. Let xt (t) be the endpoint pose of the Touch device and xb (t) be the actual endpoint pose of the Baxter robot, respectively. Then the desired endpoint pose of the Baxter robot xdb (t) can be formulated as xdb (t) = xb (ts ) + ks (xt (t) − xt (ts ))
TABLE 3. The DH parameters of the baxter robot.
(4)
where ks is the scaling factor and ts is the switching time instant. 2) The operating space mapping method is switched to the joint space mapping method. Similarly, let qt (t) be the joint position of the Touch device and qb (t) be the actual joint position of the Baxter robot, respectively. Then the desired joint position of the Baxter robot qdb (t) can be written as follows qdb (t) = qb (ts ) + (qt (t) − qt (ts )) Z t +(qt (ts ) − qb (ts )) · min 1, αdt (5) ts
where α ∈ (0, 1) is a constant and indicates the time length of the switching procedure above. B. OPERATING SPACE MAPPING METHOD
The operating space mapping method maps the pose of the endpoint of the master device to that of the slave robot. To this end, obtaining the kinematic models of these two devices is necessary. The DH parameters of the master and slave devices are shown in Table 2 and Table 3, respectively. 15360
IV. CONTROL STRUCTURE OF THE TELEOPERATION ROBOT SYSTEM
The developed control system of the teleoperation robot system, whose structure is shown in Fig. 8, integrates three parts: the workspace mapping discussed in the previous section, the joint position controller and the haptic render. VOLUME 6, 2018
Z. Chen et al.: Modular Development of Master-Slave Asymmetric Teleoperation Systems With a Novel Workspace Mapping Algorithm
Note that the last three pivot elements of Kj and Ko are specified as zero since the haptic feedback is only available for the first three joints of the Geomagic Touch device. V. EXPERIMENT
FIGURE 8. The teleoperation control structure.
The joint position controller of the Baxter robot is developed to track the desired joint command trajecotry qdb from the master device, while the haptic render is devised to enable the operator in the master site to perceive the tracking error of the slave robot indirectly through the feedback force. Typically, the following torque feedforward control law [14] is specified as the joint position controller of the baxter robot u = Mq¨ db +C(q˙ db , qdb )+g(qdb )+Kp (qdb − qb )+Kd (q˙ db − q˙ b ), (6) where M, C and g are the inertia matrix, Coriolis and centrifugal matrix and gravity matrix, respectively. qdb , q˙ db and q¨ db are the desired joint position, velocity and acceleration of the Baxter robot, respectively, while qb and q˙ b are the actual joint position and velocity of the Baxter robot, respectively. Kp and Kd are the proportional and differential diagonal feedback gain matrix, respectively. In addition, advanced control strategies like adaptive robust control [16]–[19], two timescale control [20], learning adaptive robust control [21], [22], filter-based adaptive control [23] and µ synthesis based control [24], which have been reported in many other industrial applications, can be also applied to achieve further performance improvements of the motion control in the teleoperation robot cases. The haptic render is based on the tracking error information in the slave robot site. Note that the hybrid workspace mapping method has two kinds of mapping methods, which results in two different tracking error representations. Specifically, 1) In the joint space mapping method, the tracking error is determined by the desired and actual joint angles of the Baxter robot, and thus the haptic rendering law is formulated as Fj =
Kj (qdb
− qb )
(7)
where Kj = diag{Kj1 , Kj2 , Kj3 , 0, 0, 0}. 2) In the operating space mapping method, the tracking error is obtained based on the desired and actual endpoint pose of the Baxter robot, and thus the haptic rendering law is designed as Fp = Ko (xdb − xb ) where Ko = diag{Ko1 , Ko2 , Ko3 , 0, 0, 0}. VOLUME 6, 2018
(8)
In this section, several experiments are carried out to verify our modular teleoperation platform and the proposed workspace mapping approach. Specifically, the following three experiments are compared. C1 Teleoperation task on the developed platform by using the workspace mapping method in [8] with a scaling factor of 4.5; C2 Teleoperation task on the developed platform by using the workspace mapping method in [8] with a scaling factor of 1; C3 Teleoperation task on the developed platform by using the hybrid workspace mapping method proposed in this paper with the scaling factor ks = 1 in Eq. (4). In the experiments, the endpoint tracking error of the Baxter robot is fed back to the Touch device to generate a feedback force, which can make the human operator perceive the actual tracking performance in the slave site. Specifically, the feedback force here is simply proportional to the endpoint tracking error of the Baxter robot, which indicates that the feedback force actually reflects the actual tracking performance in the slave site.
FIGURE 9. Workspace of the Touch device (Red) and the Baxter robot (Blue) for C1.
The workspace mapping results of C1, C2 and C3 are shown in Fig. 9, Fig. 10 and Fig. 11, respectively. Evidently, in C2, the mapped workspace of the Touch device only covers quite little of the actual Baxter robot workspace since a fairly conservative scaling factor is used in C2. As a result, the available motion or task range of the Baxter robot in C2 is fully limited. To deal with this issue, typically, an aggressive scaling factor can be applied instead, which is shown in C1. Compared with Fig. 10, the reachable workspace of the Baxter robot in C1 is improved greatly after applying a larger scaling factor. Furthermore, the feedback forces of C1, 15361
Z. Chen et al.: Modular Development of Master-Slave Asymmetric Teleoperation Systems With a Novel Workspace Mapping Algorithm
FIGURE 13. Endpoint trajectory tracking result for C1 (Blue: Touch device; Red: Baxter Robot).
FIGURE 10. Workspace of the Touch device (Red) and the Baxter robot (Blue) for C2.
FIGURE 14. Endpoint trajectory tracking result for C2 (Blue: Touch device; Red: Baxter Robot).
FIGURE 11. Workspace of the Touch device (Red) and the Baxter robot (Blue) for C3.
fact: an aggressive scaling factor in the workspace mapping method [8] makes the trajectory in the slave site be more sensitive to the variation of the command trajectory from the master site. As a result, the tracking performance of the slave robot degrades greatly. Therefore, covering the workspace of the slave robot and elaborate operations can not be achieved simultaneously by applying the workspace mapping method in [8].
FIGURE 12. Feedback Force (unit: N) for C1, C2 and C3.
FIGURE 15. Endpoint trajectory tracking result for C3 (Blue: Touch device; Red: Baxter Robot).
C2 and C3 are shown in Fig. 12, and the actual endpoint trajectory tracking results of C1, C2 and C3 are shown in Fig. 13, Fig. 14 and Fig. 15, respectively. From Fig. 12, Fig. 13 and Fig. 14, it is observed that the tracking performance of C1 is much poorer than that of C2 which is due to the following
In contrast, from Fig. 11, by applying the novel hybrid workspace mapping method proposed in this paper, the whole workspace of the slave robot can be reached. Furthermore, Fig. 12 and Fig. 15 indicate that an improved tracking performance is also achieved in C3. In this way, the reachable workspace for the slave robot is maximized and
15362
VOLUME 6, 2018
Z. Chen et al.: Modular Development of Master-Slave Asymmetric Teleoperation Systems With a Novel Workspace Mapping Algorithm
elaboration operations in the task space are also guaranteed, which validates the superiority of the proposed hybrid workspace mapping approach. VI. CONCLUSION
This paper develops a novel modular teleoperation robot system platform, which integrates a master Geomagic Touch device, a slave Baxter robot, ROS and MATLAB/ SIMULINK. Control system designers can benefit from the proposed platform since control algorithms can be developed in MATLAB/SIMULINK directly and efficiently. A hybrid workspace mapping method is proposed to deal with the issue of workspace discrepancies between the master and slave robots. In addition, a haptic rendering algorithm is designed to enable the operator in the master site to perceive the actual tracking error of the slave robots indirectly through the feedback force. Experimental results demonstrate the effectiveness of the proposed teleoperation system and algorithms. REFERENCES [1] W. Wei and Y. Kui, ‘‘Teleoperated manipulator for leak detection of sealed radioactive sources,’’ in Proc. IEEE Int. Conf. Robot. Autom. (ICRA), vol. 2. Apr. 2004, pp. 1682–1687. [2] W.-K. Yoon et al., ‘‘Model-based space robot teleoperation of ETS-VII manipulator,’’ IEEE Trans. Robot. Autom., vol. 20, no. 3, pp. 602–612, Jun. 2004. [3] T. Hirabayashi, J. Akizono, T. Yamamoto, H. Sakai, and H. Yano, ‘‘Teleoperation of construction machines with haptic information for underwater applications,’’ Autom. Construct., vol. 15, no. 5, pp. 563–570, 2006. [4] G. H. Ballantyne, ‘‘Robotic surgery, telerobotic surgery, telepresence, and telementoring,’’ Surg. Endoscopy, vol. 16, no. 10, pp. 1389–1402, 2002. [5] Z. Chen, Y.-J. Pan, and J. Gu, ‘‘Integrated adaptive robust control for multilateral teleoperation systems under arbitrary time delays,’’ Int. J. Robust Nonlinear Control, vol. 26, no. 12, pp. 2708–2728, Aug. 2016. [6] P. Chotiprayanakul and D. Liu, ‘‘Workspace mapping and force control for small haptic device based robot teleoperation,’’ in Proc. Int. Conf. Inf. Autom. (ICIA), 2009, pp. 1613–1618. [7] F. Conti and O. Khatib, ‘‘Spanning large workspaces using small haptic devices,’’ in Proc. 1st Joint Eurohaptics Conf. Symp. Haptic Int. Virtual Environ. Teleoperat. Syst., World Haptics, 2005, pp. 183–188. [8] Z. Ju, C. Yang, Z. Li, L. Cheng, and H. Ma, ‘‘Teleoperation of humanoid baxter robot using haptic feedback,’’ in Proc. Int. Conf. Multisensor Fusion Inf. Integr. Intell. Syst. (MFI), 2014, pp. 1–6. [9] T. Sansanayuth, I. Nilkhamhang, and K. Tungpimolrat, ‘‘Teleoperation with inverse dynamics control for phantom OMNI haptic device,’’ in Proc. SICE Annu. Conf. (SICE), 2012, pp. 2121–2126. [10] ROS. WIKI. Accessed: Nov. 2017. [Online]. Available: http://wiki. ros.org/cn [11] S. E. Salcudean, N. M. Wong, and R. L. Hollis, ‘‘Design and control of a force-reflecting teleoperation system with magnetically levitated master and wrist,’’ IEEE Trans. Robot. Autom., vol. 11, no. 6, pp. 844–858, Dec. 1995. [12] M. Mamdouh and A. A. Ramadan, ‘‘Development of a teleoperation system with a new workspace spanning technique,’’ in Proc. IEEE Int. Conf. Robot. Biomimetics (ROBIO), Dec. 2012, pp. 1570–1575. [13] A. Martinez and E. Fernández, Learning ROS for Robotics Programming. Birmingham, U.K.: Packt Publishing, 2013. [14] P. Corke, Robotics, Vision and Control: Fundamental Algorithms in MATLAB, vol. 73. Berlin, Germany: Springer, 2011. [15] L. Sciavicco and B. Siciliano, ‘‘A solution algorithm to the inverse kinematic problem for redundant manipulators,’’ IEEE J. Robot. Autom., vol. 4, no. 4, pp. 403–410, Aug. 1988. [16] J. Yao, W. Deng, and Z. Jiao, ‘‘Adaptive control of hydraulic actuators with LuGre model-based friction compensation,’’ IEEE Trans. Ind. Electron., vol. 62, no. 10, pp. 6469–6477, Oct. 2015. VOLUME 6, 2018
[17] W. Sun, Y. Zhang, Y. Huang, H. Gao, and O. Kaynak, ‘‘Transientperformance-guaranteed robust adaptive control and its application to precision motion control systems,’’ IEEE Trans. Ind. Electron., vol. 63, no. 10, pp. 6510–6518, Mar. 2016. [18] J. Yao, Z. Jiao, and D. Ma, ‘‘A practical nonlinear adaptive control of hydraulic servomechanisms with periodic-like disturbances,’’ IEEE/ASME Trans. Mechatronics, vol. 20, no. 6, pp. 2752–2760, Dec. 2015. [19] Z. Chen, B. Yao, and Q. Wang, ‘‘Accurate motion control of linear motors with adaptive robust compensation of nonlinear electromagnetic field effect,’’ IEEE/ASME Trans. Mechatronics, vol. 18, no. 3, pp. 1122–1129, Jun. 2013. [20] W. Sun, S. Tang, H. Gao, and J. Zhao, ‘‘Two time-scale tracking control of nonholonomic wheeled mobile robots,’’ IEEE Trans. Control Syst. Technol., vol. 24, no. 6, pp. 2059–2069, Nov. 2016. [21] C. Hu, Z. Wang, Y. Zhu, M. Zhang, and H. Liu, ‘‘Performance-oriented precision LARC tracking motion control of a magnetically levitated planar motor with comparative experiments,’’ IEEE Trans. Ind. Electron., vol. 63, no. 9, pp. 5763–5773, Sep. 2016. [22] C. Hu, Z. Hu, Y. Zhu, and Z. Wang, ‘‘Advanced GTCF-LARC contouring motion controller design for an industrial X–Y linear motor stage with experimental investigation,’’ IEEE Trans. Ind. Electron., vol. 64, no. 4, pp. 3308–3318, Apr. 2016. [23] W. Sun, H. Pan, and H. Gao, ‘‘Filter-based adaptive vibration control for active vehicle suspensions with electrohydraulic actuators,’’ IEEE Trans. Veh. Technol., vol. 65, no. 6, pp. 4619–4626, May 2016. [24] Z. Chen, B. Yao, and Q. Wang, ‘‘µ-synthesis-based adaptive robust control of linear motor driven stages with high-frequency dynamics: A case study,’’ IEEE/ASME Trans. Mechatronics, vol. 20, no. 3, pp. 1482–1490, Jun. 2015.
ZHENG CHEN received the B.Eng. and Ph.D. degrees in mechatronic control engineering from Zhejiang University, Hangzhou, China, in 2007 and 2012, respectively. From 2013 to 2015, he was a Post-Doctoral Researcher with the Department of Mechanical Engineering, Dalhousie University, Halifax, NS, Canada. Since 2015, he has been an Associate Professor with the Ocean College, Zhejiang University. His research interests mainly include advanced control of robotic and mechatronic system (e.g., nonlinear adaptive robust control, motion control, trajectory planning, tele-robotics, exoskeleton, mobile manipulator, precision mechatronic system, and underwater robot).
SHUIFENG YAN received the B.Eng. and M.Eng. degrees in mechatronic control engineering from Zhejiang University, Hangzhou, China, in 2014 and 2017, respectively. Since 2017, he has been an Engineer of Huawei Enterprises Telecommunication Technologies Co., Ltd.
MINGXING YUAN received the B.Eng. degree in aircraft manufacturing engineering from the Nanjing University of Aeronautics and Astronautics, Nanjing, China, in 2013. He is currently pursuing the Ph.D. degree with the College of Mechanical Engineering, Zhejiang University, Hangzhou, China. From 2016 to 2017, he was a Visiting Scholar in mechanical engineering with Purdue University, West Lafayette, IN, USA.
15363
Z. Chen et al.: Modular Development of Master-Slave Asymmetric Teleoperation Systems With a Novel Workspace Mapping Algorithm
BIN YAO received the B.Eng. degree in applied mechanics from the Beijing University of Aeronautics and Astronautics, Beijing, China, in 1987, the M.Eng. degree in electrical engineering from Nanyang Technological University, Singapore, in 1992, and the Ph.D. degree in mechanical engineering from the University of California at Berkeley, Berkeley, CA, USA, in 1996. Since 1996, he has been with the School of Mechanical Engineering, Purdue University, West Lafayette, IN, USA, where he has been a Professor since 2007 (currently on leave). He is currently a Chang Jiang Chair Professor with Zhejiang University, Hangzhou, China. He is a fellow of ASME. He was honored as a Kuang-Piu Professor in 2005 and a Changjiang Chair Professor at Zhejiang University by the Ministry of Education of China in 2010. He was a recipient of the O. Hugo Schuck Best Paper (Theory) Award from the American Automatic Control Council in 2004, the Outstanding Young Investigator Award of the American Society of Mechanical Engineers (ASME) Dynamic Systems and Control Division (DSCD) in 2007, and the Best Conference Paper Awards on Mechatronics of ASME DSCD in 2012.
15364
JINFEI HU received the B.Eng. degree in mechanical and electrical engineering from Nantong University, Nantong, China, in 2011. He is currently pursuing the Ph.D. degree with the College of Mechanical Engineering, Zhejiang University, Hangzhou, China.
VOLUME 6, 2018