BASIC STRATEGIES TO ENHANCE TELEOPERATION

PLATFORMS THROUGH THE INTERNET

J. J. Escribano, J. A. Cerrada, S. R. G

omez

Dpto. de Lenguajes y Sistemas Inform

aticos UNED

c/ Juan del rosal, 16 - 28040 Madrid (Spain)

C. Cerrada

Dpto. de Ingenier

ıa El

ectrica, Electr

onica y de Control UNED

c/ Juan del rosal, 12 - 28040 Madrid (Spain)

Keywords:

Virtual environments, teleoperation, Internet.

Abstract:

This paper shows in a schematic way the basic objectives to enhance teleoperation platforms, that have been

achieved in the PhD Thesis (Escribano, 2003). These objectives are the result of acquired experience working

with the teleoperation platform developed by our research group in teleoperation. The objectives carried out

have been formulated after analyzing the performance of the initial platform and detecting some restrictions

in several key parts of its design. Speciﬁcally, three basic improvement areas have been take into account. The

enhance objectives have been focused in aspects such as the operating system, the human-machine interface

and ﬁnally, in aspects regarding to the communication between the master and the slave. In this paper the main

detected problems are introduced as well as the chosen solutions in each case. Experimental results regarding

to the last enhancement objective are ﬁnally shown.

1 INTRODUCTION

Most of teleoperation platforms have been develo-

ped under this thinking: the more capabilities you

put, the better results you will obtain (Antal K. Be-

jczy and Schenker, 1994), (W. J. Book and Obergfell,

1996), (Luo and Chen, 1997). Various strategies to

enhance teleoperation platforms through the Internet

are shown in this paper. These strategies are inspired

in the application of low cost philosophy in both soft-

ware and hardware components. Low cost does not

mean bad performance or poor results, but it means

to invest as much money as the research need and no

more(Gomez S. R. and C., 1999).

The starting point of this work has been the previ-

ous experience acquired in the research line of tele-

operation existing in our group. As a result of this

research activity, we already had a teleoperation plat-

form. The ﬁrst step was to analyze performance ca-

pabilities of the platform in order to identify its main

limitations. Three major areas of improvement were

detected corresponding to the three enhance objec-

tives taken into account in this work. These objectives

will be shown in the next sections.

This paper is organized as follows. The capabili-

ties analysis of the previous platform is presented in

the second section. The third section shows the three

enhance objectives considered. Experimental results

of implemented solutions are presented in the fourth

section. The conclusions are shown in the last section.

2 CAPABILITIES EVALUATION

ANALYSIS

During the experimentation with this platform several

limitations have been brought forth. These constraints

have induced the research work carried out. Although

the different strategies developed to enhance this plat-

form will be described in the next section, here we

present the detected limitations. Suggested actuation

lines are also introduced.

2.1 Operating system

Human-machine interface, control system and

master-slave communication system were running

under a no real time operating system. Therefore,

critical time measurements were not reliable, making

very hard to evaluate platform performance and to

implement realistic control strategies.

242

J. Escribano J., A. Cerrada J., R. Gómez S. and Cerrada C. (2004).

BASIC STRATEGIES TO ENHANCE TELEOPERATION PLATFORMS THROUGH THE INTERNET.

In Proceedings of the Sixth International Conference on Enterprise Information Systems, pages 242-245

DOI: 10.5220/0002628102420245

 SciTePress

2.2 Human-machine interface

development tool

The main constrains of the previous human-machine

interface will be shortly described. The origin of

these limitations were located in the utilization of a

commercial software called VRT from Superscape as

developing tool for this interface. They can be su-

mmarized as follows:

• It is a commercial and closed software.

Due to this, support can disappear any time, ma-

king impossible the adaptability to any later on re-

quirement. Our proposal was to develope software

of our own allowing the addition to the system of

any new capability.

• It has no 3D information extraction capabilities

from 2D images.

The used environment did not have any special de-

vice to showing three dimensional visual informa-

tion. A solution to this problem could be adding

projected shadows to the 2D generated scenes, but

the software tool did not have this capability. It

should be very interesting to include this capability

because projected shadows allow to the operator

extracting the third dimension from a 2D image.

• The collision detection is acomplished by means of

bounding boxes.

Any object must be included inside a box (boun-

ding box). If the bounding box of an object ﬁts to

it very well, the bounding box allow to calculate

collisions well, but in the case of irregular shaped

objects, it would seem that the collisions do not ha-

ppen in the objects surface. As a solution to this

problem we suggest to use the own object surface

for collision detection.

2.3 Master slave communication

system

The communication with the slave system was be-

ing carried out thought an standard RS232 serial port.

Due to this fact the slave system must be very near

to the master system. It should be very advantageous

to implement the communication through the Inter-

net. In this way both systems could be located at any

place of the planet.

3 IMPLEMENTED STRATEGIES

From the previous study a set of enhance strategies

have been proposed and implemented in our experi-

mental platform. They can be classiﬁed in the same

three categories of the earlier analysis. This section

is devoted to describe the enhancements implemented

in each category.

• Operating system

– RTLinux (FSMLabs, ) has been selected and

used as real time operating system to substi-

tute the previous version that was developed un-

der Windows 9X. With this transformation the

teleoperation platform behaves as a synchronous

system that works with a period t controlled and

stable. In this way, platform capabilities have en-

hanced thanks to the improvement of the critical

time measurement processes, so necessary to a-

nalyze and evaluate the actual teleoperation plat-

form performance. Additionally, the use of a real

time operating system allows to include more re-

liable controllers that enhance the time response

of the operator, the slower part of all the involved

in a teleoperation system.

– A new algorithm that allows to ﬁt the data rea-

ding speed from the used Spacemouse to any pe-

riod t, in spite of the Spacemouse is only able to

provide data every 112 msec., has been imple-

mented.

• Human-Machine Interface

– By using the OpenGL library as a base, a C++

classes library that allows to make 3D scenes and

to render them in few milliseconds has been im-

plemented.

– A collision detection free library (van den

Bergen, ) has been used to implement a new al-

gorithm of maximum nearness to objects in case

of collision with the robot tool.

– The inclusion of a projected shadows algorithm

has allowed to demonstrate by means of an

experimental analysis that virtual environments

shown in monoscopic diplays allow to reach bet-

ter performance that if they are shown with stere-

ovision in virtual teleoperation platforms. This

study was presented in a more detailed way in

(J. J. Escribano, 2000). This performance en-

hancement is obtained by adding projected shad-

ows to virtual environment. An example of de-

veloped virtual environment that includes pro-

jected shadows is shown in the ﬁgure 1.

• Connection of master and slave systems

– An statistical module has been developed to ana-

lyze the Internet segment that communicates the

master system to the slave system by means of

port 7 UDP. This module allows to know the cur-

rent state of the segment and to proceed accor-

dingly.

– An Internet segment simulator that introduces

a time delay d in the receipt of packets, follo-

wing d a normal ditribution N (µ, σ) (µ and σ are

BASIC STRATEGIES TO ENHANCE TELEOPERATION PLATFORMS THROUGH THE INTERNET

243

Figure 1: Virtual environment with projected shadows

modiﬁable), and that takes into account packet

loss (also modiﬁable), has been implemented. It

has allowed to experiment with several distances

between master and slave systems.

– Several methods that try to avoid the double

problem related to communication through the

Internet (delay and packets loss) have been im-

plemented. Modiﬁcations of these methods have

been accomplished with the purpose of impro-

ving them. A comparative study between ori-

ginal methods and modiﬁed methods, similar to

the study of projected shadows, has been rea-

lised. As a result of this study it has been demon-

strated that the modiﬁed methods present impor-

tant enhancements with respect to the original

ones. Part of this study is shown in the next sec-

tion.

4 EXPERIMENTAL RESULTS

In this section it will be shown the main results of

the above mentioned comparative study concerning to

master slave communication system through the In-

ternet. Part of the obtained experimental results re-

garding to collisions measurement, backward move-

ments and times to complete a task of several meth-

ods of teleoperation through the Internet (Oboe and

Fiorini, 1998), (Brady and Tarn, 1998), (Oboe, 2001)

implemented in the PhD thesis (Escribano, 2003) are

presented here. Thirty tries for each of the considered

typical cases of connections through the Internet were

done. The mean and the standard deviation of the de-

lay for each considered case were extracted from the

table (Oboe and Fiorini, 1998) (see the table 1).

To avoid the effect of the operator acclimation to

the methods, the four cases were not tested in the

same order for each method. They were done in the

order presented in the table 2.

Next, the case 1 will be commented. In the case

Table 1: Delay parameters in typical connections through

the Internet

Case remote

machine

location

Distance

(Km)

µ(ms) σ(ms) % loss

1 Different

continent

10000 326.3 27.20 41.4

2 Different

city

150 17.20 9.74 0.80

3 Same

domain

30 8.10 5.35 0.08

4 Local 0.05 0.998 0.715 0.00

Table 2: Case order for each method in the experiments

Cases 1 2 3 4

Analitic expression 1 9 13 5

Kalman 14 6 2 10

Enhanced Kalman 11 3 7 15

Enhanced buffer 8 16 12 4

1, i.e. different continent connection, it can be ob-

served that enhanced Kalman method behaves much

better than Kalman method, both in collisions and

backwards movements (see ﬁgures 2 and 3). In the

ﬁgure 4 it can be seen that analitic expression method

detaches from the other three methods, because of its

worse behavior in completion task time. In the ﬁgure

5 it can be seen in detail that behavior of the other

three methods is similar in completion task time.

Figure 2: Collitions in case 1

As conclusion of this comparative study the me-

thods can be ordered according to their behavior.

This behavior is signiﬁcatively different when dis-

tance connection increases. In such situations, the

method that has a better behavior is enhanced buffer

method, next method is enhanced Kalman, after this

is the Kalman method and the last method is analitic

expression method. As it can be observed, it is not

necessary to use sophisticated method as a Kalman

ﬁlter in order to obtain good results in teleoperation

through the Internet.

ICEIS 2004 - HUMAN-COMPUTER INTERACTION

244

Figure 3: Backwards movements in case 1

Figure 4: Completion task times in case 1

Figure 5: Detail of completion task times in case 1

5 CONCLUSIONS

The strategies of enhancement of teleoperation plat-

forms through the Internet developed in the frame of

the PhD thesis (Escribano, 2003), have been shown

in a schematic way. These strategies are based on the

acquired skill working with the teleoperation platform

developed by our research group. Speciﬁcally, three

basic strategies have been taken into account. These

strategies are focused on the initial operating system

that was not a real time operating system, so it im-

posed hard restrictions with regard to time measure-

ments in critical processes, on the human-machine in-

terface that was very constricted to accomplish con-

trasted performance analysis, and at last on the com-

munication between master and slave systems, that

was not implemented through the Internet in the ini-

tial design. In this paper the main detected problems

are introduced as well as the chosen solutions in each

case. More explicitly, experimental results have been

shown regarding to the last enhancement strategy.

ACKNOWLEDGES

This research has been carried out under ﬁnancial

support of the project DPI2002-03999-C02-02.

REFERENCES

Antal K. Bejczy, Paolo Fiorini, W. S. K. and Schenker, P.

(1994). Toward integrated operator interface for ad-

vanced teleoperation under time-delay. In Proc. of the

IEEE/RSJ/GI International Conference on Intelligent

Robots and Systems (IROS), pages 563–570, Munich,

Germany.

Brady, K. and Tarn, T. J. (1998). Internet-based remote tele-

operation. In Proc. of the IEEE 1998 International

Conference on Robotics and Automation, pages 65–

70, Leuven, Belgium.

Escribano, J. J. (2003). Estrategias de dise

no de platafor-

mas de teleoperaci

on a trav

es de Internet. PhD thesis,

E.T.S.I Industriales. UNED.

FSMLabs. RTLinux. http://fsmlabs.com/community/.

Gomez S. R., Carro J., M. E. E. J. J. and C., C. (1999). A

design strategy for low cost experimental telerobotics

platforms. In Proc. IEEE Int. Workshop on Robot and

Human Interaction, pages 273–278, Pisa, Italy.

J. J. Escribano, C. Cerrada, S. R. G. J. A. C. (2000). Stereo

vision vs. projected shadows in performance evalua-

tion of teleoperated systems. In Proc. IEEE Int. Work-

shop on Robot and Human Interaction, Osaka, Japan.

Luo, R. C. and Chen, T. M. (1997). Remote supervisory

control of a sensor based mobile robot via internet. In

Proc. IROS’97, pages 1163–1168.

Oboe, R. (2001). Web-interfaced, force-reﬂecting teleoper-

ation systems. IEEE Trans. on Industrial Electronics.,

48(6):1257–1265.

Oboe, R. and Fiorini, P. (1998). A design and control envi-

ronment for internet-based telerobotics. The Interna-

tional Journal of Robotics Research, 17(4):433–449.

van den Bergen, G. FreeSolid: Software Library for Inter-

ference Detection. http://www.win.tue.nl/

gino/solid/.

W. J. Book, H. Lane, L. J. L. D. P. M. and Obergfell, K.

(1996). A novel teleoperated long-reach manipula-

tor tesbed and its remote capabilities via the internet.

In Proc. of the IEEE 1996 International Conference

on Robotics and Automation, pages 1036–1041, Min-

neapolis, Minnesota.

BASIC STRATEGIES TO ENHANCE TELEOPERATION PLATFORMS THROUGH THE INTERNET

245