28.08.2014 Views

Fite, Shao, and Goldfarb.pdf - Vanderbilt University

Fite, Shao, and Goldfarb.pdf - Vanderbilt University

Fite, Shao, and Goldfarb.pdf - Vanderbilt University

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

620 IEEE TRANSACTIONS ON ROBOTICS AND AUTOMATION, VOL. 20, NO. 3, JUNE 2004<br />

Loop Shaping for Transparency <strong>and</strong> Stability<br />

Robustness in Bilateral Telemanipulation<br />

Kevin B. <strong>Fite</strong>, Liang<strong>Shao</strong>, <strong>and</strong> Michael <strong>Goldfarb</strong><br />

Abstract—This paper presents <strong>and</strong> experimentally demonstrates a control<br />

methodology that provides transparency <strong>and</strong> stability robustness in<br />

bilateral telemanipulator systems. The approach is based upon a previously<br />

published method that structures the human-manipulators-environment<br />

system in a manner that enables the application of frequency-domain<br />

loop-shaping methods. This paper reformulates the human-manipulator<br />

interaction described in the previously published work,<strong>and</strong> experimentally<br />

demonstrates the approach on a single degree-of-freedom telemanipulation<br />

system. Experimental measurements indicate significant improvements<br />

offered by the method in both the stability robustness <strong>and</strong> transparency<br />

of the human-manipulators-environment system. Finally,experimental<br />

results are presented that demonstrate the robustness in the transparency<br />

to significant changes in the environment dynamics.<br />

Index Terms—Bilateral,telemanipulation,teleoperation,transparency.<br />

I. INTRODUCTION<br />

A bilateral telemanipulator enables human interaction with environments<br />

that are remote, hazardous, or otherwise inaccessible to<br />

direct human contact. The performance of such a system is often<br />

characterized by its ability to present the undistorted dynamics of the<br />

environment to the human operator, a characteristic termed “transparency.”<br />

The ability to do so is compromised by the (closed-loop)<br />

dynamics of the master <strong>and</strong> slave manipulators, which distort the<br />

dynamics of the environment as perceived by the human operator.<br />

Additionally, the human, telemanipulator pair, <strong>and</strong> environment form<br />

an interacting system that, given conditions of high environment<br />

impedance or high loop gain, is susceptible to instability. Specifically,<br />

though each component of the system can itself be robustly stable,<br />

the accumulation of phase from multiple components can render<br />

the human-manipulators-environment system unstable. A common<br />

objective of bilateral telemanipulation is, therefore, to design a<br />

controller that provides transparency <strong>and</strong> stability robustness of the<br />

telemanipulator system in the presence of significant variation in<br />

the dynamics of the environment <strong>and</strong> human operator. It should be<br />

noted, however, that some researchers have proposed methods in<br />

which the environment impedance is intentionally dynamically altered<br />

to remedy the kinematic similarity issues that arise when scaling<br />

dynamic environments [1], [2].<br />

II. PRIOR WORK<br />

Several researchers have investigated aspects of transparency <strong>and</strong><br />

stability in telemanipulation, includingHannaford [3], Yokokohji <strong>and</strong><br />

Yoshikawa [4], Lawrence [5], Hashstrudi-Zaad <strong>and</strong> Salcudean [6], Anderson<br />

<strong>and</strong> Spong[7], Niemeyer <strong>and</strong> Slotine [8], Yoshikawa <strong>and</strong> Ueda<br />

[9], <strong>and</strong> Munir <strong>and</strong> Book [10]. These <strong>and</strong> related prior works have<br />

incorporated two-port network representations, as introduced into the<br />

telemanipulation literature in [3], for the analysis of telemanipulator<br />

Manuscript received February 3, 2003. This paper was recommended for publication<br />

by Associate Editor P. Dupont <strong>and</strong> Editor I. Walker upon evaluation of<br />

the reviewers’ comments.<br />

K. <strong>Fite</strong> <strong>and</strong> M. <strong>Goldfarb</strong> are with the Department of Mechanical Engineering,<br />

V<strong>and</strong>erbilt <strong>University</strong>, Nashville, TN 37235 USA (e-mail: fitekb@vuse.v<strong>and</strong>erbilt.edu;<br />

goldfarb@vuse.v<strong>and</strong>erbilt.edu).<br />

L. <strong>Shao</strong> was with the Department of Mechanical Engineering, V<strong>and</strong>erbilt <strong>University</strong>,<br />

Nashville, TN 37235 USA.<br />

Digital Object Identifier 10.1109/TRA.2004.825474<br />

Fig. 1.<br />

Two-channel bilateral telemanipulation.<br />

transparency <strong>and</strong> the design of associated filters to provide transparent<br />

performance. Specifically, controllers are designed to yield a two-port<br />

telemanipulator hybrid parameter matrix that approximates the identity.<br />

Havingaddressed the performance objective, the stability of the<br />

two-port is addressed by incorporatingpassivity concepts. Such concepts,<br />

however, are conservative, <strong>and</strong> as such, compromise system performance<br />

(i.e., transparency) more than necessary.<br />

This paper was in part motivated by the fact that prior publications<br />

on this topic seem to uniformly incorporate network-based concepts to<br />

address transparency <strong>and</strong> passivity-based concepts to address stability,<br />

the former necessitatingthe latter. Rather than use hybrid two-port<br />

network theory, the control architecture presented in this paper<br />

addresses the performance <strong>and</strong> stability of a two-channel telemanipulation<br />

system from a frequency-domain perspective. Specifically,<br />

the human-manipulators-environment system is structured such<br />

that a single loop is formed, to which st<strong>and</strong>ard frequency-domain<br />

loop-shapingmethods can be applied. Lawrence et al. incorporated<br />

similar notions in assessingthe stability <strong>and</strong> transparency of the<br />

haptic renderingof virtual objects [11]. Such an approach enables a<br />

less conservative control design, <strong>and</strong> therefore, better stability <strong>and</strong><br />

transparency properties. This approach was initially described for<br />

purposes of bilateral telemanipulation by <strong>Fite</strong> et al. [12]. This paper<br />

significantly alters the formulation of the method published in [12],<br />

primarily by reformulatingthe human-manipulator interaction, <strong>and</strong><br />

experimentally demonstrates on a single degree-of-freedom (DOF)<br />

telemanipulation system both the stability robustness <strong>and</strong> performance<br />

robustness offered by the proposed method.<br />

Finally, it should be noted that this paper does not explicitly consider<br />

or treat the presence of time delay in the teleoperator communication<br />

channels. Such delay is commonly present in remote teleoperation<br />

systems, but is generally not significant in local teleoperation<br />

systems, such as scaled teleoperation systems for dexterity enhancement<br />

or glove-box teleoperation systems for purposes of biological or<br />

radiological isolation.<br />

III. FORMATION OF HUMAN-MANIPULATORS-ENVIRONMENT LOOP<br />

Fig. 1 depicts the general notion of two-channel bilateral telemanipulation,<br />

in which a human operator interacts with a force-controlled<br />

master manipulator, which is in turn coupled to a position-controlled<br />

slave manipulator interactingwith an environment. The two subsystems<br />

are coupled through scaled motion <strong>and</strong> force communication<br />

channels, where C 1 <strong>and</strong> C 2 represent the motion <strong>and</strong> force-scaling<br />

gains, respectively. The motion comm<strong>and</strong> from the master/human<br />

subsystem, Xh, is the combined effect of human voluntary motion <strong>and</strong><br />

the “feedthrough” motion from the teleoperator loop. The latter results<br />

from a comm<strong>and</strong>ed motion Xh that is filtered by the slave/environment<br />

dynamics <strong>and</strong>, in turn, generates a force at the master, Fh, which, in<br />

turn, acts upon the human admittance <strong>and</strong> results in a component of the<br />

comm<strong>and</strong>ed motion. Instability will result when the phase lagin the<br />

teleoperator system is such that this force is 180 out of phase with the<br />

comm<strong>and</strong>ed motion, <strong>and</strong> the loop gain is at least one. In order to assess<br />

the transparency <strong>and</strong> relative stability of the loop, the teleoperator<br />

system must be restructured so that the human voluntary input, which<br />

1042-296X/04$20.00 © 2004 IEEE


IEEE TRANSACTIONS ON ROBOTICS AND AUTOMATION, VOL. 20, NO. 3, JUNE 2004 621<br />

Fig. 2.<br />

Model of master/human dynamics.<br />

Fig. 3. Parsing the master/human dynamics. (a) Schematic of closed-loop<br />

force-controlled master manipulator interactingwith human. (b) Schematic of<br />

fully parsed voluntary <strong>and</strong> feedthrough effects.<br />

is essentially an exogenous input (i.e., not part of the feedback loop),<br />

is parsed from the teleoperator feedback loop. Consider, for example,<br />

a linear single-DOF master/human subsystem, which can be modeled<br />

as shown in Fig. 2 <strong>and</strong> described in [13]. In this model, the master<br />

manipulator is considered a simple mass-damper system (b m ;m m )<br />

that is kinematically coupled to the human arm. The arm is modeled as<br />

a mass-spring-damper system (m h ;b 2;k 2) with an additional stiffness<br />

<strong>and</strong> damping (b 1 ;k 1 ) at the human/master interface (i.e., compliance<br />

in the human grip). Human voluntary input x hv is comm<strong>and</strong>ed directly<br />

into the base of the arm mass-spring-damper system, <strong>and</strong> is therefore<br />

filtered by these dynamics before resultingin motion of the master<br />

manipulator, a notion that is consistent with prior studies in human<br />

motor control [14]. The iconic model clearly indicates that slave<br />

motion comm<strong>and</strong> x h can result from either the voluntary human input<br />

x hv or the feedthrough force F h . In order to express the teleoperator<br />

system in terms of a classical loop with the human voluntary input<br />

as an exogenous comm<strong>and</strong>, the multiple-input single-output (MISO)<br />

human arm system depicted in Fig. 2 must be restructured as follows.<br />

Assumingthat the master closed-loop force controller is described by<br />

the transfer function C m , the master/human dynamics can be written<br />

in the block diagram form shown in Fig. 3(a), where G 1; G 2, <strong>and</strong> G 3<br />

are transfer function matrices given by<br />

where<br />

A =<br />

G 1 = 0<br />

G 2 = 0 0 0<br />

b 2 s + k 2<br />

m m<br />

0 0<br />

C m<br />

m m<br />

T<br />

T<br />

(1)<br />

(2)<br />

G 3 = C(sI 0 A) 01 (3)<br />

0 1 0 0<br />

0(k +k ) 0(b +b ) k b<br />

m<br />

m m m<br />

0 0 0 1<br />

k<br />

b 0k 0(b +b )<br />

m<br />

m m m<br />

C =[0k 1 0b 1 k 1 b 1] (5)<br />

(4)<br />

Fig. 4. Slave/environment dynamics. (a) Schematic of closed-loop<br />

motion-controlled slave manipulator interactingwith environment impedance.<br />

(b) Restructuringof interaction, indicatingdependence of G on Z . (c) Use<br />

of local feedback of environment interaction force to decouple G from Z .<br />

(d) Schematic of resultingdecoupled dynamics.<br />

<strong>and</strong> where Y h is the admittance of the human operator, given by<br />

Y h = sX h<br />

F h<br />

[m h s 2 +(b 2 + b 1)s +(k 2 + k 1)]s<br />

=<br />

: (6)<br />

m h b 1s 3 +(m h k 1 + b 1b 2)s 2 +(k 1b 2 + k 2b 1)s + k 1k 2<br />

All parameters in (1)–(6) are as defined in Fig. 2. The block diagram<br />

of Fig. 3(a) can be rearranged until the respective paths of the<br />

human voluntary input X hv <strong>and</strong> the feedthrough force F h contributing<br />

to the comm<strong>and</strong> motion X h are separated completely, as shown in<br />

Fig. 3(b). Specifically, the transfer function describing the force component<br />

actingon the human admittance resultingfrom human voluntary<br />

input is given by G hv , <strong>and</strong> the transfer function describingthe<br />

force component actingon the human admittance resultingfrom the<br />

feedthrough force is given by G m. Both transfer functions can be computed<br />

from (1)–(5) <strong>and</strong> from the expressions given in Fig. 3.<br />

The slave/environment dynamics depicted in Fig. 1 can be<br />

represented by the equivalent schematics of Fig. 4(a) <strong>and</strong> (b),<br />

both of which clearly indicate the dynamic couplingbetween the<br />

slave <strong>and</strong> environment. In the figures, Z e;Y s, <strong>and</strong> C s represent<br />

the environment impedance, slave manipulator admittance, <strong>and</strong><br />

position controller, respectively, <strong>and</strong> G ~ s represents the closed-loop<br />

single-input, single-output (SISO) slave transfer function, which is<br />

clearly a function of the environment impedance. Reconfiguring the<br />

human-manipulators-environment system into a classical feedback<br />

loop, as described by Figs. 3(b) <strong>and</strong> 4(b), enables application of<br />

classical control techniques to address the transparency <strong>and</strong> stability<br />

of the system, as initially described in [12]. The proposed control<br />

approach incorporates two modifications to the two-channel bilateral


622 IEEE TRANSACTIONS ON ROBOTICS AND AUTOMATION, VOL. 20, NO. 3, JUNE 2004<br />

Fig. 5. Two-channel telemanipulation architecture. Solid arrowheads<br />

represent signal interaction, whereas hollow ones represent physical<br />

interaction.<br />

structure shown in Fig. 1 to address the transparency <strong>and</strong> stability robustness<br />

of the system. First, the closed-loop slave dynamics includes<br />

local feedback of the interaction force between the slave manipulator<br />

<strong>and</strong> environment, as shown in Fig. 4(c). This component effectively<br />

compensates for the coupled slave/environment interaction, yielding<br />

closed-loop slave dynamics G s independent of the environment, as<br />

depicted in Fig. 4(d). Second, the control architecture includes a<br />

loop-shapingcompensator, G c , operatingon the comm<strong>and</strong> motion X h<br />

that enables frequency-domain manipulation of both the transparency<br />

<strong>and</strong> stability robustness properties of the teleoperation loop. The<br />

resultingteleoperator loop can be modeled as shown in Fig. 5, where<br />

G m;G hv , <strong>and</strong> G s are transfer functions as defined by Figs. 3 <strong>and</strong> 4.<br />

Assumingunity scalinggains <strong>and</strong> neglectingcommunication<br />

channel time delay, the transmitted impedance, which is the<br />

impedance of the teleoperator loop seen by the human operator, is<br />

given by<br />

Z t =<br />

F m<br />

sX h<br />

= G cG sG mZ e: (7)<br />

The transparency transfer function, defined as the ratio of the transmitted<br />

impedance to the environment impedance, is written as<br />

G t = G c G s G m : (8)<br />

The performance-control design objective is to choose a compensator<br />

G c that yields a transparency transfer function G t with a magnitude<br />

of unity <strong>and</strong> phase of zero within some desired b<strong>and</strong>width of<br />

operation. Note that without the previously mentioned local force feedback<br />

around the slave manipulator, the transfer function G s would be a<br />

function of the environment impedance, Z e , <strong>and</strong> as a result, the transparency<br />

transfer function G t would similarly change with varying environmental<br />

dynamics. Inclusion of the local slave force-feedback loop,<br />

however, renders G s , <strong>and</strong> thus G t , independent of changes in the environment<br />

dynamics (see Fig. 4).<br />

The stability robustness of the teleoperator loop is assessed by rearranging<br />

the loop into a Nyquist-like unity-feedback structure, enabled<br />

by parsingthe human voluntary force from the feedthrough force, as<br />

previously described. The resultingopen-loop transfer function governingthe<br />

stability robustness is given by<br />

G = 0G c G s G m Z e Y h : (9)<br />

The stability robustness of the loop is thus easily addressed by adding<br />

a desired amount of phase at the open-loop gain crossover frequency<br />

with the compensator G c. The design approach is thus to use the magnitude<br />

characteristics of G c to address the system transparency, <strong>and</strong><br />

use the phase characteristics to address the system stability. Though<br />

Fig. 6. Top view of slave manipulator interacting with an environment<br />

stiffness.<br />

several possibilities for such compensator design exist, a reasonably<br />

general solution is of a lead-lag compensator of the form<br />

N i s +1<br />

G c = kc5i=1 (10)<br />

i i s +1<br />

where the parameters k c ;N; i , <strong>and</strong> i are used to shape the magnitude<br />

<strong>and</strong> phase characteristics of the compensator.<br />

IV. IMPLEMENTATION<br />

To verify the proposed control approach, the loop-shapingtechnique<br />

was experimentally implemented on a single-DOF telemanipulation<br />

system. Both the master <strong>and</strong> slave were kinematically single-DOF prismatic<br />

manipulators, with actuation provided by a DC brushed servomotor<br />

(PMI N12M4T) transformed to linear motion via a rack <strong>and</strong><br />

pinion. A rotary potentiometer (Midori CPP-45B) was used for rotor<br />

position measurement. A rotational inertia was additionally mounted<br />

to each servomotor to represent the inertia of a typical manipulator.<br />

The master manipulator incorporated a h<strong>and</strong>le mounted on the end of a<br />

cantilever beam coupled to the translatingrack to provide an interface<br />

with the human operator. Strain gauges mounted on the cantilever measured<br />

the interaction forces occurringbetween the human operator <strong>and</strong><br />

the manipulator. The slave manipulator incorporated a cantilever beam<br />

that connected its endpoint to a pair of springs supported by a shaft<br />

mounted parallel to the linear motion. The springs imposed a bidirectional<br />

stiffness in series with the slave motion, providinga simple yet<br />

challenging environment with which to assess the teleoperative performance.<br />

Similar to the master, the slave incorporated strain gauges on<br />

the cantilever beam to measure the interaction forces between the slave<br />

<strong>and</strong> environment. Each manipulator was capable of exertinga maximum<br />

continuous force of approximately 45 N through a workspace<br />

of approximately 65 cm. The slave <strong>and</strong> master manipulators are pictured<br />

in Figs. 6 <strong>and</strong> 7, respectively. The control architecture was implemented<br />

with the real-time interface provided by MATLAB/Simulink<br />

(The MathWorks, Inc.) at a samplingrate of 1 kHz.<br />

A. Master <strong>and</strong> Slave Manipulator Control<br />

The nominal dynamics of the slave manipulator result from the rotational<br />

inertia previously described <strong>and</strong> the (assumed) viscous friction<br />

of the motor brushes <strong>and</strong> bearings, both transformed into translational<br />

equivalents via the rack <strong>and</strong> pinion. The nominal (translational)


IEEE TRANSACTIONS ON ROBOTICS AND AUTOMATION, VOL. 20, NO. 3, JUNE 2004 623<br />

Fig. 8. Transparency transfer function for the teleoperator system without<br />

loop compensation, demonstratinga 63-dB transparency b<strong>and</strong>width of<br />

approximately 1.4 Hz.<br />

Fig. 7.<br />

Side view of human operator gripping the master manipulator.<br />

mass <strong>and</strong> dampingof the slave manipulator were experimentally determined<br />

to be m s =4:5 kg<strong>and</strong> b s =80Ns/m, respectively. A proportional-derivative<br />

compensator, attenuated at high frequencies with a<br />

first-order filter, was used for the slave endpoint position control, such<br />

that<br />

where<br />

C s =<br />

k ds s + k ps<br />

s+1<br />

k ps =2360N 1 m<br />

k ds = 260 Ns/m<br />

=0:01 s:<br />

(11)<br />

Similar to the slave manipulator, the nominal dynamics of the master<br />

result also from the rotational inertia <strong>and</strong> (assumed) viscous friction of<br />

the motor, both transformed into translational equivalents via the rack<br />

<strong>and</strong> pinion. The inertia <strong>and</strong> dampingassociated with the master manipulator<br />

were experimentally determined to be m m = 12.25 kg<strong>and</strong><br />

b m = 100 Ns/m, respectively. A proportional controller with a gain of<br />

k pm = 8 N/N was used to provide endpoint force control of the master<br />

manipulator. Finally, the nominal environment stiffness used in the experiments<br />

was 750 N1m.<br />

B. Experimental Measurement of Transparency <strong>and</strong> Stability<br />

Robustness<br />

The transparency of the teleoperative system was assessed by<br />

measuringthe experimental frequency response of the transparency<br />

transfer function. Specifically, the human operator excited the<br />

closed-loop system with a semir<strong>and</strong>om excitation. Measurements of<br />

the motion <strong>and</strong> resultingimposed force, occurringat the interface<br />

between the master <strong>and</strong> human operator, were made over a 30-s time<br />

interval. The experimental frequency response was obtained from the<br />

measured data usingthe expression<br />

Z t;exp (j!) =<br />

F h(j!)<br />

sX h (j!) = 8VF(j!) (12)<br />

8 VV(j!)<br />

where 8 VF(j!) is the cross-power spectral density between the motion<br />

input <strong>and</strong> the force output, <strong>and</strong> 8 VV (j!) is the power spectral<br />

density of the motion input. The transparency transfer function was<br />

then obtained by dividingthe experimental measure of the transmitted<br />

impedance by the actual environment impedance.<br />

The stability margins were experimentally obtained by breaking the<br />

loop at the motion comm<strong>and</strong> to the slave <strong>and</strong> introducingsinusoidal ex-<br />

citation to measure the open-loop (time-based) response of the system.<br />

The gain margin of the loop was determined by measuring the sinusoidal<br />

magnitude of the loop transfer function for excitation at the frequency<br />

where the output human motion lagged the input by 180 . The<br />

phase margin was found by measuring the amount of lag between the<br />

input <strong>and</strong> output for excitation at the frequency for which the sinusoidal<br />

magnitude of the output equals that of the input.<br />

C. Transparency <strong>and</strong> Stability Robustness of the Uncompensated<br />

System<br />

A baseline experiment incorporated the structure shown in Fig. 5,<br />

but without the loop-shapingcompensator, G c . Fig. 8 shows the experimentally<br />

measured frequency-response plot of the transparency for<br />

the uncompensated loop. The transparency exhibits a DC offset that<br />

effectively attenuates the perceived magnitude of the environment. Ignoringthe<br />

presence of the offset, the loop attains a 63-dB transparency<br />

b<strong>and</strong>width of approximately 1.4 Hz. The gain <strong>and</strong> phase margins were<br />

measured as previously described to be 6.7 dB <strong>and</strong> approximately 37 ,<br />

respectively.<br />

D. Transparency <strong>and</strong> Stability Robustness of the Compensated System<br />

Usingthe previously described experimental measures of transparency<br />

<strong>and</strong> stability robustness, a loop-shapingcompensator, G c ,as<br />

described by (10) was designed with the following parameters:<br />

k c =1:12<br />

N =1<br />

1 =0:133<br />

1 =0:028 s:<br />

Inclusion of the compensator as shown in Fig. 5 rectified the DC offset<br />

exhibited by the uncompensated system <strong>and</strong> increased the transparency<br />

b<strong>and</strong>width to approximately 3 Hz, as shown in Fig. 9. The compensator<br />

additionally increased the stability robustness of the teleoperation<br />

system, providingmeasured gain <strong>and</strong> phase margins of 10 dB <strong>and</strong><br />

49 , respectively. Addition of the compensator therefore increased the<br />

transparency b<strong>and</strong>width by more than a factor of two, <strong>and</strong> additionally<br />

increased the gain <strong>and</strong> phase margins by factors of 50% <strong>and</strong> 38%, respectively.<br />

E. Performance Robustness<br />

As previously discussed, the local feedback of the interaction force<br />

between the slave manipulator <strong>and</strong> environment eliminates the dependence<br />

of the transparency on the environment dynamics, <strong>and</strong> thus pro-


624 IEEE TRANSACTIONS ON ROBOTICS AND AUTOMATION, VOL. 20, NO. 3, JUNE 2004<br />

V. CONCLUSION<br />

A control methodology that provides transparency <strong>and</strong> stability<br />

robustness in bilateral telemanipulator systems was experimentally<br />

demonstrated on a single-DOF telemanipulation system. Specifically,<br />

the approach was shown to provide significant improvement in both<br />

the transparency <strong>and</strong> stability robustness of the teleoperator system,<br />

<strong>and</strong> was shown to maintain a consistent transparency b<strong>and</strong>width<br />

independent of variation in the environment.<br />

Fig. 9. Transparency transfer function for the teleoperator system with<br />

loop compensation, demonstratinga 63-dB transparency b<strong>and</strong>width of<br />

approximately 3.0 Hz.<br />

Fig. 10. Transparency transfer function of the compensated system for an<br />

environment stiffness 375, 750, <strong>and</strong> 2250 N1m.<br />

vides robustness to changes in the environment impedance. Such robustness<br />

was experimentally validated by repeatingthe transparency<br />

measurements with the same compensator, but with a decreased environment<br />

stiffness of 375 N1m <strong>and</strong> an increased environment stiffness of<br />

2250 N1m, respectively. The results of these measurements are shown<br />

in Fig. 10, which shows the transparency transfer function for the nominal,<br />

decreased, <strong>and</strong> increased environment stiffnesses. The compensated<br />

teleoperation system clearly maintains the nominally measured<br />

transparency for significant changes in the impedance of the environment<br />

with which the telemanipulator interacts.<br />

REFERENCES<br />

[1] J. E. Colgate, “Robust impedance-shaping telemanipulation,” IEEE<br />

Trans. Robot. Automat., vol. 9, pp. 374–384, Aug. 1993.<br />

[2] K. Kaneko, H. Tokashiki, K. Tanie, <strong>and</strong> K. Komoriya, “Impedance<br />

shapingbased on force feedback bilateral control in macro-micro teleoperation<br />

system,” in Proc. IEEE Int. Conf. Robotics <strong>and</strong> Automation,<br />

1997, pp. 710–717.<br />

[3] B. Hannaford, “A design framework for teleoperators with kinesthetic<br />

feedback,” IEEE Trans. Robot. Automat., vol. 5, pp. 426–434, Aug.<br />

1989.<br />

[4] Y. Yokokohji <strong>and</strong> T. Yoshikawa, “Bilateral control of master-slave<br />

manipulators for ideal kinesthetic coupling,” in Proc. IEEE Int. Conf.<br />

Robotics <strong>and</strong> Automation, 1992, pp. 849–858.<br />

[5] D. A. Lawrence, “Stability <strong>and</strong> transparency in bilateral telemanipulation,”<br />

IEEE Trans. Robot. Automat., vol. 9, pp. 624–637, Oct. 1993.<br />

[6] K. Hashtrudi-Zaad <strong>and</strong> S. E. Salcudean, “On the use of local force feedback<br />

for transparent teleoperation,” in Proc. IEEE Int. Conf. Robotics<br />

<strong>and</strong> Automation, 1999, pp. 1863–1869.<br />

[7] R. J. Anderson <strong>and</strong> M. W. Spong, “Bilateral control of teleoperators with<br />

time delay,” IEEE Trans. Automat. Contr., vol. 34, pp. 494–501, May<br />

1989.<br />

[8] G. Niemeyer <strong>and</strong> J. E. Slotine, “Stable adaptive teleoperation,” IEEE J.<br />

Ocean. Eng., vol. 16, pp. 152–162, Jan. 1992.<br />

[9] T. Yoshikawa <strong>and</strong> J. Ueda, “Analysis <strong>and</strong> control of master-slave systems<br />

with time delay,” in Proc. IEEE Int. Conf. Intelligent Robots <strong>and</strong><br />

Systems, 1996, pp. 1366–1373.<br />

[10] S. Munir <strong>and</strong> W. J. Book, “Internet-based teleoperation usingwave<br />

variables with prediction,” IEEE/ASME Trans. Mechatron., vol. 7, pp.<br />

124–133, Feb. 2002.<br />

[11] D. A. Lawrence, L. Y. Pao, M. A. Salada, <strong>and</strong> A. M. Dougherty, “Quantitative<br />

experimental analysis of transparency <strong>and</strong> stability in haptic interfaces,”<br />

in Pro. ASME Int. Mechanical Engineering Congr. Expo., 1996,<br />

pp. 441–449.<br />

[12] K. <strong>Fite</strong>, J. Speich, <strong>and</strong> M. <strong>Goldfarb</strong>, “Transparency <strong>and</strong> stability in twochannel<br />

bilateral telemanipulation,” J. Dynam. Syst., Meas., Contr., vol.<br />

123, pp. 400–407, Mar. 2001.<br />

[13] J. E. Speich, S. Liang, <strong>and</strong> M. <strong>Goldfarb</strong>, “An experimental h<strong>and</strong>/arm<br />

model for human interaction with a telemanipulation system,” in<br />

Proc. ASME Int. Mechanical Engineering Congr. Expo., 2001, Paper<br />

IMECE2001/DSC-24617.<br />

[14] N. Hogan, “Mechanical impedance of single- <strong>and</strong> multi-articular systems,”<br />

in Multiple Muscle Systems: Biomechanics <strong>and</strong> Movement Organization,<br />

J. M. Winters <strong>and</strong> S. L.-Y. Woo, Eds. New York: Springer-<br />

Verlag, 1990.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!