A Survey of Telesensation and Teleoperation Technology with Virtual Reality and Force Reflection Capabilities

Pattaraphol Batsomboon, Sabri Tosunoglu
Florida International University
Department of Mechanical Engineering
Miami, Florida 33199
E-mail: tosun@fiu.edu

Daniel W. Repperger
Armstrong Laboratory, AL/CFBA
Human Sensory Feedback Laboratory
Wright Patterson Air Force Base
Dayton, Ohio 45433
E-mail: d.repperger@ieee.org

Abstract - A telesensation system implies the ultimate goal of teleoperation. Such a system provides the operator with the sensational feelings of the remote site as if he/she were working in the actual environment. Thus, the performance of the system is greatly improved. This paper mainly reviews some of the existing units of the three most important elements in a telesensation system. These units are force reflecting manual controllers, virtual reality units and an advanced operator interface. The feel of touch is provided by the force-reflecting manual controller, while the virtual reality unit gives the operator the 3-D view of the working environment. The telesensation system, also implemented with the advanced operator interface, allows the operator to easily control, model, plan and simulate remote systems. These features have proven to be the most significant feedback of telesensation systems.

INTRODUCTION

Teleoperation is a general term that refers to a human-machine remote control system. The main function of a teleoperation system is to assist the operator to perform complex and uncertain tasks in hazardous environments such as space, nuclear reactors and under-water operations. For instance, robotic technologies have been used to inspect, maintain and service the nuclear power plants [28]. As a result, the radiation exposure of workers at the plants has been reduced to the lowest possible level [22].

In the past decades, teleoperator systems have been developed to the point where the human operator is able to perceive the feelings as if he/she were in the actual environment. Such a system is also referred to as telepresence. A more recent goal in the development of teleoperation is telesensation. The word "telesensation" has been used to describe telecommunication systems such as teleconferencing where people from different remote locations in the real world are able to hold a meeting or work cooperatively in the same artificial world. Such a system combines the use of computer vision, computer graphics, virtual reality (VR) and telecommunications [26, 27].

In the field of robotics, the term "telesensation" implies the advanced teleoperator system can provide the operator with sensation feedback by employing the five senses (if possible). A flexible programming environment and better integration of human and computer capabilities highlight the advanced teleoperation technology [23]. The telesensation system, depicted in Figure 1, integrates the use of an advanced operator interface, virtual reality unit, force-reflecting manual controller and sensor-based manipulator to provide the feel of presence at the remote site [2]. Thus, with the skilled operator and the sophisticated systems, the tasks can be accomplished with relative ease and fidelity.

This paper reviews some the three most important elements in the telesensation system: a force-reflecting manual controller, a virtual reality unit and an operator interface. These units make the telesensation system superior to traditional teleoperator systems by providing the operator with the sensation of a 3-D visual feedback through the use of a VR unit, the feel of touch from a force-reflecting manual controller and an enhanced control with the advanced operator interface.

A VR system not only provides the operator with the 3-D view of remote site, but also sound feedback, voice input and motion tracking [4]. With a virtual reality-based telesensation system, the operator can guide the remote system through voice commands from the VR unit or by the use of a force-reflecting manual controller. Force feedback is one of the most significant elements of feedback information for a telesensation system. It is fundamental for mechanical support, sense of balance and a feel of touching real objects. It conveys information that is essential in many activities such as training, design analysis and hazardous-environment task simulations [17]. These kinds of feedback information significantly enhance productivity of the remote system.

In addition, an advanced operator interface design enables the operator to better control the remote system. The rapid development of computer graphics technologies makes it possible to integrate high-fidelity real-time graphics simulation/animation into advanced teleoperations [11]. A friendly graphical user interface can extensively assist the operator in accomplishing tasks in all phases of work [10]. Thus, it improves the efficiency, reliability and safety of the system.

Figure 1. A Pictorial Representation of Components in a Telesensation System

FORCE-REFLECTING MANUAL CONTROLLERS

A force-reflecting manual controller (FRMC), or joystick, is one of the devices that can be used to control remote systems in teleoperation. A joystick is often a better control device than other available options such as a mouse, switchbox, keyboard or touch-screen input because the operator identifies better with the task [8]. However, to apply the concept of telesensation to a conventional teleoperation system, the joystick should be able to reflect forces experienced at the remote site. Such a system is known as a force-reflecting manual controller. While the input motion moves the remote system, forces experienced by the system are reflected to the manual controller so that the operator feels the forces acting on the system [1]. The FRMC provides a virtual force [21] which enhances the realism of the virtual reality-based telesensation system; thus, the performance of the operator is greatly improved.

Force-Reflecting Mechanical Design

The architecture of force-reflecting manual controller mechanisms is classified either as serial or parallel structures. Either of these two architectures has certain advantages and disadvantages when applied to the telesensation systems. For instance, while the serial structure provides a larger workspace, parallel mechanisms tend to be more compact. The relative merits of these two categories of mechanisms are given in [1, 5]. Some of the serial- and parallel-structured manual controllers are reviewed below.

Serial-Structured Design Force-Reflecting Manual Controllers

Teleoperator System SM-229

In 1977, Teleoperator System Corporation developed a bilateral force-reflecting servo master-slave manipulator called SM-229. It had seven degrees of freedom with a 3.7 m³ workspace. It was designed to serve the requirements of a variety of new installations in nuclear plants. Conceivably, the SM-229 was the first member of a family of force-reflecting electric master-slave manipulators designed to be produced and maintainable [18].

Bilateral Force-Reflecting 6-DOF Manual Controller

In 1980, Jet Propulsion Laboratory (JPL) and Stanford Research Institute (SRI) developed a universal, bilateral force-reflecting 6-DOF manual controller [9]. The design effort was to minimize friction, backlash and inertia at the handgrip. The system used the cable/pulley-based counter-balancing and power transmitting mechanism which was capable of generating a force up to 35 oz at the handgrip. In addition, a counterbalance assembly was included in the system to reduce gravitational effects.

Handyman

The Handyman electrohydraulic manipulator was developed by General Electric company in 1985 [9]. The system includes articulated fingers and an exoskeleton force-reflecting master arm. However, the Handyman did not prove to be practical because of its large size.

Remote Manipulator System (RMS)

In 1970's, NASA developed the Remote Manipulator System (RMS) for the space shuttle [9]. The system uses two 3-DOF hand controllers: one for translational motion and the other for rotational motion of the end effector. The controllers do not reflect forces experienced by the RMS. The RMS primarily uses a resolved unilateral rate control of the individual joints.

Maintenance System (M-2)

The Model M-2 Maintenance System was developed at the Oak Ridge National Laboratory (ORNL) in an effort to improve remote manipulation technology for nuclear fuel reprocessing and other remote applications [18]. The system consists of two force-reflecting master controllers for two servomanipulator arms, television viewing, lighting and auxiliary lifting capabilities. The touch-screen system was used as an interface between the operator and the remote manipulators. The features include force ratio selection, camera/lighting control and system status diagnostics.

Advanced Servomanipulator (ASM) System

This remote maintainable force-reflecting servomanipulator system was also developed at ORNL [18]. The main objective of the ASM was to use it in reprocessing maintenance which required reliability, radiation tolerance and corrosion resistance. The ASM uses eight remotely replaceable module types where each module weighs less than 23 kg so that they can be carried by another ASM. The servomanipulator uses gear systems which provide a payload capacity of 23 kg. One of the main differences of the ASM from the traditional system is the anthropomorphic (elbow-down) geometry. As a result, the ASM was restricted in its applications.

Kraft KMC-9100

The Kraft KMC-9100 force-reflecting hand controller is produced by Martin Marietta/Kraft. This compact system has six degrees of freedom and is able to reflect forces up to 5 pounds when fully extended. It is kinematically similar to the human arm which creates an intuitive relationship between the operator's movements and those translated to the manipulator. It is intended to be used for the Flight Telerobotic Servicer (FTS).

Cybernet System PER-Force

Two versions of PER-Force, 3 DOF and 6 DOF, have been produced by the Cybernet System Corporation. The 3-DOF version consists of three 30 oz-in brushless DC motors. It is capable of reflecting a maximum force of 9 pounds and yet the joystick unit weighs only 4.5 pounds. The 6-DOF version (Figure 2) incorporates three linear axes with the position resolution of 0.0003" per location and three revolute axes with 1/90 degree or 40 seconds position resolution. The unit was originally designed for the Space Station. Both versions can be controlled by IBM, VME or Macintosh-compatible computers.

Figure 2. PER-Force 6-DOF Force-Reflecting Manual Controller

Parallel-Structured Design Force-Reflecting Manual Controllers

Stewart Platform

The Stewart platform was first introduced by Stewart [9]. It has six degrees of freedom and uses all six actuated prismatic joints. However, because the prismatic actuators are usually not backdrivable, this type of design cannot be used for tasks requiring compliance of the manipulator.

Nine-String Force-Reflecting Six-DOF Manual Controller

Based on the design of the Stewart platform, a nine-string six-DOF manual controller has been developed at the University of Texas at Austin [16]. It is capable of reflecting forces up to ten pounds by using nine actuators to control nine string tensions. In addition, three constant-pressure air cylinders are used to provide constant compression forces where the strings cannot provide the force needed. The workspace has no singularities, and because the motion of each string is measured by a potentiometer, the computational burden is reduced. On the other hand, the system is rather bulky and has high friction from the pneumatic cylinders.

Three-DOF Spherical Shoulder Manual Controller

A three-DOF spherical shoulder manual controller has also been developed at the University of Texas. This system has almost the same features as the nine-string controller except that it has only three degrees of freedom. As in most parallel-structured mechanisms, the spherical shoulder allows the location of heavy actuators on the ground; thus, increasing the payload capacity. However, because each actuator is integrated with a harmonic drive system with a 60:1 gear ratio, the system exhibits high magnitudes of friction, backlash and inertia forces due to the high gear-ratio reducers in the actuator modules [9].

FIU Three-DOF Manual Controller

The conceptual design of a three-DOF force-reflecting manual controller, which was developed at Florida International University, is shown in Figure 3. The system utilizes a direct drive setup which eliminates the need of intermediate transmission elements such as gears or belts. As a result, it has zero backlash and virtually almost no friction. The system consists of three powerful, small rare-earth permanent magnet brushless DC motors which provide a maximum reflected force of five pounds. The design is expected to be one of the most compact three-DOF manual controllers.

Figure 3. FIU 3-DOF Force-Reflecting Manual Controller Design

VIRTUAL REALITY UNIT

A Virtual Reality (VR) unit is a visual device, similar to a helmet, that enables a person to perceive and interact with a virtual environment as if it were real [19]. In telesensation systems, the VR unit provides the view of the remote site as the operator turns or tilts his/her head which corresponds to a remote system camera. A standard TV monitor is not able to provide the operator with a sense of a 3-D view of the working environment. Thus, the feeling of presence and sensation is reduced which results in poor performance [29].

One of the biggest advantages of having a VR unit is that the actual manipulator can be removed from the training loop while the operator is being trained. This training approach can be quite beneficial because the robot cannot be damaged by any mistreatment by an operator, and the operator should feel more comfortable during the training exercise. In addition, training scenarios can be set up easily. These scenarios usually include an emergency situation procedure training. Should the actual emergency occur, the operator would be much better in handling the situation [19].

The Current State-of-the-Art VR Systems

Vision

VR systems usually include stereo vision. Two types of displays used in lower cost systems are Liquid Crystal Displays (LCD) and CRT. The CRT display areas are usually small with high light output while flat-panel LCD displays have low weight and optional color, but with poor resolution and relatively low light output. Thus, CRTs are more preferable in display design with folded optical paths.

In terms of complexity and realism, VR units have similar objectives of visual image generation to those in aircraft simulation. However, faster scene changes are required in the VR systems as a result of user head movements. VR units must be able to provide effective visual simulation which requires computational complexity in order to eliminate hidden lines and produce effective perspective. In addition, computational speed must be fast enough to provide the operator with acceptable scene update rates. Currently, one of the most powerful and popular work stations is produced by Silicon Graphics.

The image generator produces an output by receiving the signal from the measurement of head movements which are measured optically, acoustically, mechanically or magnetically. One of the most popular systems is the Polhemus Spasyn system. This technique requires only small sensors to be mounted on the head and is insensitive to most interference. However, one of the biggest disadvantages of this system is that the accuracy is affected by metal, and the environments must be mapped extensively.

Audio

An audio system is one of the most important elements in VR units. Spatially distinct sounds are important attributes of a convincing virtual reality. One approach to produce a successful virtual 3-D sound is to apply a mathematical function called Head-Related Transfer Function or HRTF. The HRTF relates to an individual's ear shape, but generalized HRTFs have been successfully created that work for most people. Research has shown that perceptual errors can cause problems, such as sounds behind the head that are perceived as if they were in the front of the head. These types of problems cannot be solved even when generalized HRTFs are used [30].

One of the most efficient VR systems, built at the Armstrong Aerospace Medical Research Laboratory, is called the Visually Coupled Airborne Systems Simulator (VCASS). The system uses miniature CRTs as image resources to produce high-resolution displays. VCASS provides a binocular field of view of 120 degrees horizontal and 60 degrees vertical. It has high bandwidth video amplifiers, programmable analog circuits for pre-distorting the images and many other features. Other proficient VR units have been produced by Honeywell for use in the "Falcon Eye," the F-16 night attack system, the Apache attack helicopter and in other various British research projects.

Extensive teleoperation research projects using the VR technology have been developed at NASA's AMES Research Project [25]; the JPL to control remotely deployed robots [3]; the Automation and Robotics (A&R) Division at the Johnson Space Center (JSC) for telepresence research, robotics and extravehicular activities (EVA) analysis and training [20]; the Advanced Controls Manipulation Laboratory (ACML) at Sandia National Laboratories for waste remediation technologies; the University of Tennessee at Knoxville Mobile-Manipulator Robotics Research (M2R2); and the Oak Ridge National Laboratory (ORNL)-Advanced Servo Manipulator (ASM)-based Decontamination and Decommissioning (D&D) [29].

Some of the commercial VR units available in the market are briefly reviewed below. These include the Visual Immersion Module, VIM, Liquid Image MRG2.2, Virtual Research System VR4 and Personal Use Stereoscopic Haptic PUSH virtual reality systems.

Figure 4. Visual Immersion Module (VIM)

Visual Immersion Module (VIM)

The Visual Immersion Module (VIM), as shown in Figure 4, is produced by Kaiser Electro-Optics, Inc. There are two models available: 500HRpv and 1000HRpv. The 500HRpv employs two full-color, 0.7" Active Matrix LCD (AMLCD) display with a resolution of 180,000 pixels per LCD. The field of view (FOV) is approximately 50 and weighs only 26 oz. The 1000HRpv provides with four full-color displays instead of two and has a vertical FOV of 30 and 100 horizontally. Both provide built-in Sennheiser 2 channel stereo headphones.

Liquid Image MRG2.2

The MRG2.2 (Figure 5) is one of the various VR units manufactured by the Liquid Image Corporation. The unit employs a single, full color AMLCD display type with a resolution of 240 720 pixels per eye. It has a FOV of 84 horizontally and 65 vertically. Internally mounted microphones and positional tracker are two of the many options that can be incorporated.

Figure 5. Liquid Image MRG2.2

Virtual Research Systems VR4

The VR4 (Figure 6) is manufactured by Virtual Research Systems, Inc. This light-weight unit (33 oz.) has a field of view of 60 diagonal at 100% overlap and 67 at 85% overlap. Each eye has a resolution of 742 230 pixels which is equivalent to 56,887 triads. The VR4 employs a 1.3" diagonal AMLCD display type and Sennheiser HD440 digitally compatible headphones.

Figure 6. Virtual Research Systems VR4

Personal Use Stereoscopic Haptic (PUSH)

The Personal Use Stereoscopic Haptic (PUSH), as shown in Figure 7, is a desktop immersive display produced by Fakespace. The unit provides an adjustable field of view of 30 to 140 with a resolution of 1280 1024 triads per eye. It has 6-DOF control with 3-DOF PUSH interface. A test has been conducted by Fakespace to examine the unit. Users ranging from experts to novice were asked to navigate through a virtual test environment. With PUSH technology, the company claims that the users were able to navigate in a large virtual space or zoom in to see the objects from different viewpoints.

Figure 7. Personal Use Stereoscopic Haptic (PUSH)

For the purpose of implementing a VR unit in a telesensation system, although any of the systems reviewed above might be used, either the VIM 1000HRpv or VR4 would be more suitable. Both systems provide built-in stereo headphones and a good range of field of view with fine resolution. Their light weight and reasonable costs make them more attractive.

OPERATOR INTERFACE SYSTEM

Force-Reflecting Controller Design

In a force-reflecting teleoperation system, two control modes, position and force, must be implemented in the control loops. Whitney notes that when a force is exerted by the end-effector on the environment, a force control mode is perpendicular to the environment and the position control mode can be exerted to the environment tangentially [31].

Figure 8 depicts the concept of an open-loop control in a force-reflecting teleoperator system. This scheme may not provide satisfactory results because the control system is unmodelled for dynamics, friction, backlash, delay, etc., between the teleoperator and remote system. As a result, the system may become unstable.

Figure 8. Open-Loop Control of a Telesensation System

A more sophisticated control scheme, the closed-loop control, is shown in Figure 9. The output of the remote system is directly fed back to the input of the manual controller while the local feedback loops still remain in the control loop. This closed-loop control accounts for dynamics, friction, backlash, delay and perhaps, beam damping which is an important parameter for the long-reach manipulator [7].

Figure 9. Closed-Loop Control of a Telesensation System

Some of the well-known control methods used for a force-reflecting teleoperation system are kinesthetic force feedback, shared compliant control [12] and "telemonitoring" sensory feedback [15]. These methods provide force feedback except that kinesthetic force feedback has a stiffer system. In kinesthetic force feedback, the forces sensed by the remote manipulator are fed back and are reflected through the operator's manual controller, whereas in a shared compliant control, the human operator shares the control task with the autonomous compliant control of the remote manipulator. In the telemonitoring control mode, the system consists of a position control with a position error-based force reflection and remote site compliance [14]. Its control mode basically is under a shared compliant control but has telemonitoring force feedback. The comparison of the performance of these control methods are given in [1].

Supervisory Interface Software

The general structure of the software structure is presented in Figure 10. Some of the desirable aspects of the interface software are described below.

Figure 10. Telesensation System Computer Graphical Interface

Menu-Driven Interface. In a telesensation system or in an advanced teleoperator system, a friendly graphical user interface (GUI) that supports a menu-driven windows environment is a very important element. This type of sophisticated software enhances the performance of the operator to accomplish the required tasks more efficiently.

Graphics Window. By providing graphics window displays in the telesensation system, the operator is able to interface with the remote system by means of pictorial communications [10]. An advanced graphical operator interface can be employed in all three phases of teleoperation servicing and inspection: off-line task analysis and planning, operator training and on-line task execution [13]. The amount of detail and accuracy in the displayed model depend on the fidelity of the system. The fidelity also includes the smoothness of an animated remote manipulator and the update time between the operator's motions and the simulation's motions [11]. By using such graphics displays, the mobile robotic systems can be designed, developed and operated in a more efficient and reliable fashion.

Joystick Control Software

Position and Force Scaling. This type of joystick control software allows the user to input the scaling values of position and force reflection between the joystick and the manipulator. In the position mode, as the operator moves the joystick a certain distance, the manipulator would move according to the set scaling value. In the force mode, the operator is able to input the scaling value of the reflected force at the manual controller. For instance, when the robot experiences a certain amount of force, the magnitude of the reflected force is scaled by the operator. This is especially important to prevent fatigue. This flexibility makes it possible to accomplish various tasks at any level of delicacy.

Rereferencing. As the operator moves the joystick, the manipulator moves a certain distance. However, in order to move the manipulator further, the operator must define a reference point so that he/she can make other joystick movements. This is accomplished by temporarily suspending the joystick/remote system connection, moving the joystick to a desirable reference position within its workspace and then establishing the connection with the remote system. This is achieved by placing a connect/disconnect toggle switch on the joystick.

Joystick Mechanism Input-Output and Force Equations. The joystick input-output motion equations and force-reflection equations must be derived and executed by special software. This software does the forward and inverse kinematic analysis of the manual controller and remote manipulator. These equations should be written as efficiently as possible so that they can be used in real-time computer control.

VR Control Software

One of the main advantages in implementing VR control software in a telesensation system is that VR is capable of moving the human-robot interface to a new intuitive and user-friendly level [6]. The interaction between human and computer is much more perceptive than that of using a mouse to change the view in a flat screen display. In addition, operators become familiar to the environment rapidly and require little instruction in how to use the system. In designing the VR control software, a main requirement is that the tracking of the user's head must be accurate in terms of position and orientation.

Latency and Update Rate. Aside from the accurate tracking, the prime considerations of the system designer are minimizing latency and maximizing the update rate. The software must be able to modify images rapidly and keep up with the user's head movements [24]. Both parameters indicate whether a VR system will serve as a useful device or frustrate the user [30].

VR Input Capability. The interaction between the operator and the VR system can be made through an immersive stereo viewer and voice input. An example of such a system has been developed at ACML for application in the clean-up of massive Underground Storage Tanks (UST). This system uses audio feedback to continuously guide the operator and to provide command confirmation [19].

CONCLUSIONS

A telesensation system provides the operator with the intuitive and sensational feelings of the remote site as if he/she were at the actual environment. The feedback information improves the system performance, reliability and safety. In this paper, major components of telesensation systems are reviewed. These include force-reflecting manual controllers, virtual reality units and interface systems. Various manual controllers that use either serial or parallel mechanisms are reviewed. Virtual reality units which provide 3-D sensation of the simulated remote site are also introduced. Finally, the structure of an interface system is presented. This is characterized by a graphical interface for video/graphics/VR inputs, a friendly environment for user inputs and easy access to open- or closed-loop controllers for position and force control.

ACKNOWLEDGEMENTS

This work is partially funded by the U.S. Air Force, Air Force Materiel Command (Grant No: FY8990-95-00094). The authors gratefully acknowledge this support. They also thank Ernesto Romero for his assistance in rendering some of the figures.

REFERENCES

[1] Batsomboon, P., and Tosunoglu, S., "A Review of Teleoperation and Telesensation Systems," 1996 Florida Conference on Recent Advances in Robotics, Florida Atlantic University, Boca Raton, Florida, April 11-12, 1996.

[2] Batsomboon, P., Tosunoglu S., and Repperger, D. W., "Component Technologies for Virtual Reality-Based Telesensation Systems," The Third ASME Biennial Joint Conference on Engineering Systems Design & Analysis, ASME ESDA'96, Montpellier, France, Vol. 2, pp. 229-236, July 1-4, 1996.

[3] Bejczy, A. K., "Sensors, Controls, and Man-Machine Interface for Advanced Teleoperation," Science, Vol. 208, No. 4450, pp. 1327-1335, 1980.

[4] Christopher, C., and Stansfield, S., "Interactive Graphical Model Building using Telepresence and Virtual Reality," Proceedings of the IEEE International Conference Robotics and Automation, Vol. 2, pp. 1436-1439, 1994.

[5] Conklin, W., and Tosunoglu, S., "Conceptual Design of a Universal Bilateral Manual Controller," 1996 Florida Conference on Recent Advances in Robotics, Florida Atlantic University, Boca Raton, Florida, April 11-12, 1996.

[6] Davies, B. R., "Remediating Hazardous Waste Robotically Using a High-Level Control System and Real-time Sensors," Proceedings of the SPIE International Symposium on Optical Tools for Manufacturing and Advanced Automation, 1993.

[7] Jansen, J. F., Burks, B. L., Babcock, S. M., Kress, R. L., and Hamel, W. R., "Long-Reach Manipulation for Waste Storage Tank Remediation," Modeling and Control of Compliant and Rigid Motion Systems, Vol. 31, pp. 67-73, 1991.

[8] Johnsen, E. G., and Corliss, W. R., Human Factors Applications in Teleoperator Design and Operation, John Wiley & Sons, Inc, New York, 1971.

[9] Kim, W., "A Study on the Design and Operation of Force Feedback Controllers," Ph.D. Dissertation, University of Texas at Austin, January 1991.

[10] Kim, W. S., "Graphical Operator Interface for Space Telerobotics," Proceedings of the IEEE International Conference Robotics and Automation, Vol. 3, pp. 761-768, 1993.

[11] Kim, W. S., and Bejczy, A. K., "Graphics Displays for Operator Aid in Telemanipulation," Proceedings of the IEEE International Conference on Systems, Man and Cybernetics, Vol. 2, pp. 1059-1067, 1991.

[12] Kim, W. S., Hannaford, B., and Bejczy, A. K., "Force Reflection and Shared Compliant Control in Operating Telemanipulators with Time Delay," IEEE Transactions on Robotics and Automation, Vol. 8, No. 2, pp. 176-185, April 1992.

[13] Kim, W. S., Schenker, P. S., Bejczy, A. K., and Hayati, S., "Advanced Graphics Interfaces for Telerobotic Servicing and Inspection," International Conference on Intelligent Robots and Systems, pp. 303-309, 1993.

[14] Lee S., Bekey, G. A., and Bejczy, A. K., "Computer Control of Space-borne Teleoperators with Sensory Feedback," Proceedings of the IEEE International Conference Robotics and Automation, pp. 205-214, 1985.

[15] Lee, S., and Lee, H. S., "Modeling Design and Evaluation of Advanced Teleoperator Control Systems with Short Time Delay," IEEE Transactions on Robotics and Automation, Vol. 9, No. 5, pp. 607-623, October 1993.

[16] Lindemann, R., Tosunoglu, S., and Tesar, D., "Construction and Demonstration of a Nine-String, Six-Degree-of-Freedom Force-Reflecting Joystick for Telemanipulation," Final Report to NASA Johnson Space Center, Grant No. NAG9-188, Department of Mechanical Engineering, University of Texas, Austin, Texas, October 1987.

[17] McNeely, W., Burdea, G. C., Hannford, B., Hirose, M., Jacobsen, S., Salisbury, K., and Tachi, S., "Whither force feedback?," IEEE Annual Virtual Reality International Symposium, pp. 226-227, 1995.

[18] Martin, H. L., and Kuban, D. P., Teleoperated Robotics in Hostile Environments, Robotics International of SME, Michigan, 1985.

[19] Miner, N. E., and Stansfield, S. A., "An Interactive Virtual Reality Simulation System for Robot Control and Operator Training," Proceedings of the IEEE International Conference Robotics and Automation, Vol. 2, pp. 1428-1435, 1994.

[20] Read, D. A., Ruta, K. J., and Bell, B. N., "Application of Virtual Reality to Space Robotics Operations," ASCE Specialty Conference on Robotics for Challenging Environments Albuquerque, pp. 19-27, 1994.

[21] Repperger, D. W., Phillips C. A., and Chelette, T. L., "Study on Spatially Induced 'Virtual Force' with an Information Theoretic Investigation of Human Performance," IEEE Transactions on Systems, Man and Cybernetics, Vol. 25, No. 10, pp. 1392-1404, 1995.

[22] Rochelear, D. N., and Crane, C. D. III, "Development of a Graphical Interface for Robotic Operation in a Hazardous environment," Proceedings of the IEEE International Conference on Systems, Man and Cybernetics, Vol. 2, pp. 1077-1081, 1991.

[23] Schenker, P. S., Bejczy, A. K., Kim, W. S., and Lee, S., "Advanced Man-Machine Interface and Control Architecture for Dexterous Teleoperation," Ocean Technologies and Opportunities in the Pacific for the 90's, Vol. 3, pp. 1506-1525, 1991.

[24] Sheridan, T. B., "Teleoperation, Telerobotics and Telepresence: A Progress Report," Control Engineering Practice, Vol. 3, No. 2, pp. 205-214, 1995.

[25] Stark, L., et al., "Telerobotics: Display, Control, and Communication Problems," IEEE Journal of Robotics and Automation, Vol. 3, No. 2, pp. 205-214, 1987.

[26] Terashima, N., "Telesensation - Distributed Interactive Virtual Reality - Overview and Prospects," IFIP Transactions A: Computer Science and Technology, No. A-51, pp. 49-59, 1994.

[27] Terashima, N., "VR and Telesensation: Novel Disciplines for the Next Century," IEEE Symposium on Emerging Technologies & Factory Automation, pp. 22-28, 1994.

[28] Tosunoglu, S., and Hamel, W. R., "Robotics Technology Assessment for Decontamination and Dismantle Tasks," Report to the U.S. Department of Energy, Grant No. DE-FG02-86NE37966, September 1994.

[29] Trivedi, M. M., and Chen, C., "Developing Telerobotic Systems Using Virtual Reality Concepts," Proceedings of the IEEE/RSJ International Conference on Intelligent Robots and Systems, pp. 352-359, 1993.

[30] Vince, J., Virtual Reality Systems, ACM Press, New York, 1995.

[31] Whitney, D. E., "Force Feedback Control of Manipulator Fine Motions," ASME Transaction Journal of Dynamic Systems, Measurement, and Control, pp. 91-97, June 1977.