Nowadays, one of the possibilities of performing inspections of electrical power lines is using visual inspections from video records taken from a camera situated on a commercial helicopter. This task is very dangerous because the flight must be done at very low altitude and close to the cables. It is also expensive because it is necessary to use a commercial helicopter and at least two qualified operators. This paper describes the system architecture and vehicle control of the prototype that is being developed by the robotic group of DISAM at “Universidad Politécnica de Madrid” for unmanned aerial inspection of electrical power lines. Artificial system vision is needed to obtain the information for the vehicle guidance by automatic detection of the cable lines and generating 3D coordinates. Navigation is performed using an inertial measurement unit to obtain the attitude of the craft. The information from the vision system is used to correct the helicopter trajectories. Helicopters exhibit highly non-linear systems with strongly coupled modes. They are very difficult to model and control. In this paper, a model of the minihelicopter is adjusted and controlled using a Linear Quadratic Gaussian (LQG) controller for pitch and roll movements and a PID one for yaw

Virginia Tech has developed a novel two-wheeled vehicle termed the biplanar bicycle. This vehicle differs from the traditional bicycle in that both wheels rotate about a common axis, thus orienting the wheels side by side instead of front to back. Drive motors generate the propulsive torque by acting against a central reaction mass (the vehicle body) suspended from the axle. This vehicle offers several advantages over traditional wheeled vehicles. Among these are inherent stability, self-righting ability, and excellent mobility in two degrees of freedom (the ground plane). The physical architecture makes the vehicle exceedingly nimble through a wide range of speeds. Further, the entire mass of this vehicle is distributed between the two drive wheels, thus maximizing the available frictional driving force. The vehicle dynamics resulting from the system geometry affords very low slip and good acceleration. These advantages, coupled with a simple mechanical design, make the biplanar bicycle a good candidate for use as a platform vehicle in autonomous applications. Some potential applications include rail line inspection, landmine/UXO detection, and planetary exploration. This paper presents results from prototype testing of the biplanar bicycle. Various issues concerning sensors and control strategies unique to this vehicle are examined. While the conceptual basis of this vehicle may be counterintuitive, the design is fundamentally sound and viable for autonomous vehicle applications.

The Man Portable Robotic System (MPRS) project objective was to build and deliver hardened robotic systems to the U.S. Army's 10th Mountain Division in Fort Drum, New York. The systems, specifically designed for tunnel and sewer reconnaissance, were equipped with visual and audio sensors that allowed the Army engineers to detect trip wires and booby traps before personnel entered a potentially hostile environment. The greatest challenges for the project stemmed from the users three main requirements: 1) man-portable (lightweight and small), 2) waterproof (not just water-resistant), and 3) soldier proof(highly rugged and reliable). The MPRS systems were, of course, plagued by the usual problems in robotics: limited battery power (run-time) and limited communications range. At the Army's request, the systems incorporated no autonomous functionality; however, MPRS did integrate several state-of-the-art components, including a fully digital video system. This paper discusses specific challenges encountered and lessons learned by the MPRS team during recent tunnel and sewer reconnaissance testing at three sites in 2000: Fort Drum (New York), Fort Leonard Wood (Missouri), and Fort Polk (Louisiana).

The Savannah River Site’s (SRS) Environmental Restoration Division (ERD) is in the process of reconditioning and decommissioning potentially contaminated facilities. Many of these facilities require inspection prior to remediation tasks. Remote inspections are necessary to reduce hazards and ensure the health and safety of workers and the public. To aid in ERD’s efforts, the Remote and Specialty Equipment Systems (RSES) section developed a remote inspection vehicle referred to as the Manhole Access Pipecrawler System (MAPS). The MAPS is comprised of a battery-powered mobile vehicle, a portable control console, a cable reel enclosing 300’ of tethered cable, and a deployment/retrieval system. The components used in MAPS are predominantly standard off-the-shelf items to reduce fabrication costs. Parts were chosen based on size and durability to satisfy SRS operating conditions.

Worker safety is ofparamount importance in industries where harsh working environments are the norm. The research being conducted by this group aims to take workers out ofharm's way by creating an automated control system for fleets of intelligent machines that will do the dirty work, while the humans oversee the system to ensure proper operation. The focus of the research is not on machine intelligence, but rather on the system that will control the fleet in unstructured or semistructured locales. Mine environments are used as a target for this research, which is broken into a few main sections. The first section deals with dynamically creating task schedules for the vehicles based on possibly changing environmental conditions. The second section deals with resource sharing between multiple vehicles, especially the sharing ofroadways to ensure operational safety. This is done using Petri net data structures and theory. Thirdly, since the machines may not be able to independently cope with obstacles they encounter, human intervention capability is required. Fourthly, for human operators to make sense of the system's overall state and requests, development of a human-machine interface is necessary. Experiments have been conducted which demonstrate the successful use ofthis framework to control two model-size intelligent machines given one shared resource, namely a two-road intersection. In the future the group intends on imposing greater loads on the system (i.e. more vehicles and shared resources), and on integrating more complex human intervention capabilities, finer vehicle control, and improved system state monitoring.

Regardless of the level of autonomy provided by the individual robots, teams of mobile robots will always require human command and control for tasking, monitoring, error remediation, and coordination. Multiple mobile robot systems are hard real-time multi-agent systems and are often employed in hazardous or inaccessible environments. Thus, it is necessary that human-robot interfaces (HRIs) be developed and tailored specifically to multiple mobile robot systems. These HRIs will be different from single-robot control interfaces because of the heterogeneity of the robots, the multi-tasking nature of the application, and the divided attention demands on the operator. In addition to the requisite command and control functions of the unit (e.g., a graphical user interface, communications network management, and sensor data acquisition channels), the HRI will require efficient multi-agent planning and tasking interfaces and a feedback display and execution monitoring system able to fuse diverse data. In this paper we will discuss the characteristics of multirobot HRIs and present data that should guide the design of such interfaces. In particular, we will discuss the design of multi-modal command/feedback interfaces, alert-driven presentation managers, and graphical user interfaces. Experiments and ongoing work from the DARPA Tactical Mobile Robot (TMR) program will be presented that indicate current progress in the design and implementation of such systems.

We describe techniques for coordinating the actions of large numbers of small-scale robots to achieve useful large-scale results in surveillance, reconnaissance, hazard detection, and path finding. Inspired by the chemical markers used by insects for communication and coordination, we exploit the notion of a “virtual pheromone,” implemented using simple beacons and directional sensors mounted on each robot. Virtual pheromones facilitate simple communication and coordination and require little on-board processing. Our virtual pheromone technique enables a robot collective to become a distributed computing mesh embedded within the environment, while simultaneously acting as a physical embodiment of the user interface. This leads to notions of world embedded computation and world embedded interfaces that provide new ways to think about robot colonies.

In the extreme environments posed by war fighting, fire fighting, and nuclear accident response, the cost of direct human exposure is levied in terms of injury and death. Robotic alternatives must address effective operations while removing humans from danger. This is profoundly challenging, as extreme environments inflict cumulative performance damage on exposed robotic agents. Sensing and perception are among the most vulnerable components. We present a distributed robotic system that enables autonomous reconnaissance and mapping in urban structures using teams of robots. Robot teams scout remote sites, maintain operational tempos, and successfully execute tasks, principally the construction of 3-D Maps, despite multiple agent failures. Using an economic model of agent interaction based on a free market architecture, a virtual platform (a robot colony) is synthesized where task execution does not directly depend on individual agents within the colony.

This paper proposes a conceptual design for a distributed system of very simple robots capable of performing a useful realworld mission such as mapping the interior of a building overnight with a swarm of possibly hundreds of cockroach-sized robots. Presentation of this system concept follows an initial discussion of strategies for developing distributed robotic systems. Success is dependent on making good decisions in selecting appropriate applications, in system design, and in executing the system development process. Each robot includes basic mobility, crude odometry, contact or near-contact object/obstacle detection sensors, an omnidirectional beacon (probably IR), and a beacon detection sensor that can simultaneously detect multiple beacons on other robots and measure the bearing of each to less than one degree. Beacon triangulation (combined with knowledge of some baseline distance) allows the determination of the position of any robot (and any object next to it) relative to the others. Occlusion of a robot’s beacon indicates the presence of an intervening object, while lack of occlusion identifies a “ray” of free space. Clever deployment of large numbers of robots will permit mapping of walls and objects, and can also support the detection and localization of intruders moving within the space. The object sensor can be very short range and, therefore, hopefully very simple and cheap, perhaps implemented as an array of whiskers. The beacon sensor is more complex, but it can be completely tuned to the detection of the cooperating beacon. Thus, the robots need tackle no perception tasks whatsoever. The system development begins with the implementation of an initial baseline system, in which each robot is controlled by a central coordinating element via a high bandwidth communications link. This initial effort will develop and validate behaviors and algorithms, assess sensitivity to sensor error, determine communications and processing requirements, and generally expedite the system’s evolution into a fully distributed system.

Holometrics, Inc. produces a family of patented 3-D Vision System products capable of significantly enhancing the operational effectiveness of law enforcement remotely operated, mobile robots. We have demonstrated this capability during a joint program we have with the San Diego Metro Arson Strike Team, a division ofthe San Diego Fire Department. The MAST unit mission is to provide a capability to respond to bomb threats, hostage situations, earthquakes, etc., using a remotely controlled robot to reduce exposing their personnel to dangerous situations. All ofthe MAST personnel involved in the joint program gave the 3-D system extremely high marks in terms ofits utility in improving their operational capabilities. This paper summarizes the technical and operational characteristics ofHolometrics 3-D vision systems. Particular emphasis is placed on describing those technical features that enhance mobile robot applications in explosive material handlindisposal, inspecting buildings that have been damaged by an earthquake or explosion, defining areas or buildings possibly involved in the illegal deployment ofweapons ofmass destruction, providing crucial information inside of buildings involving hostage or terrorist activities, and many others. The 3D imageiy of a large scene can be captured in just 2 — 3 seconds. The image is immediately stored thus preserving perishable scene conditions before changes occur — e.g., lighting, weather, removal ofdead or injured, structural collapse, etc. This technology is acceptable by the courts since the underlying theory for these systems is tested, they have been subjected to peer review, error rates are known and understood, and the technology is accepted by the scientific community.

Radar offers advantages as a robotic perception modality because it is not as vulnerable to the vacuum, dust, fog, rain, snow and light conditions found in construction, mining, agricultural and planetary-exploration environments. However radar has shortcomings such as a large footprint, sidelobes, specularity effects and limited range resolution—all of which result in poor environment maps. Evidence grids are a flexible and powerful probabilistic method for fusing multiple sensor observations. Sensor models exist for interpreting the range readings of sonar, laser and stereo. However, these existing sensor models do not work with radar because it provides amplitude values for many points downrange. In addition, radar has significant echo signal-to-noise variations between observations as well as limited downrange resolution. This paper presents the development of a radar sensor model, which can fuse amplitude-vector sensor data into an evidence grid. A study of radar phenomena and of frequency-modulated continuous-wave signal processing suggests rules for signal interpretation. The sensor model uses these interpretation rules and captures the volumetric beam geometry. The results include a three-dimensional map of an outdoor scene. This work is a step towards building high fidelity maps to be used in mobile robot navigation, obstacle avoidance and tool deployment under all visibility conditions.

A differentially steered three-wheeled vehicle has proven to be an effective platform for outdoor navigation. Many applications for this vehicle configuration, including planetary exploration and landmine/UXO location, require accurate localization. In spite of known problems, odometry, also called dead reckoning, remains one of the least expensive and most popular methods for localization. This paper presents the results of an investigation into the benefits of instrumenting the rear caster wheel to supplement the drive wheel encoders in odometry. A linear observer is used to fuse the data between the drive wheel encoders and the caster data. This method can also be extended using the standard form of the Kalman filter to allow for noise. Improvements in position estimation in the face of common problems such as slip and dimensional errors are quantified.

This paper proposes a method for a robot to pick up human voice clearly and remotely. The method utilizes microphones array. By setting gain and delay of each microphone properly, it enables to form "acoustic focus" at the desired location. The proposed method is intended to be used as a speech signal input for human interactive mobile robot. To confirm feasibility of the proposed idea, the authors conduct simulation and simple experiments. The simulation result supports the feasibility. As for the experiment, although the array is so simple that the number of microphones is only three and signal processing of the array is "offline", we can confirm that a sensitivity peak of the array appears at the desired location. Both results show possibility to realize "acoustic focus" at the desired location. Thus these results support feasibility of the proposed idea.

Virginia Tech is currently developing a new autonomous vehicle as a research platform. This vehicle is being used to investigate techniques in autonomous landmine/UXO detection. In addition, it was entered in the 2000 Intelligent Ground Vehicle Competition. This vehicle senses its surroundings using two (non-stereo) color CCD cameras, a SICK laser range finder, and wheel encoders. The cameras give a color representation of the area in front of the vehicle; while the laser range finder provides range data for obstacles in a 180-degree arc in front of the vehicle. Encoder feedback is used to determine position and velocity of the vehicle. This paper presents the techniques used to fuse this diverse and asynchronous data into a useful representation. The software architecture, which allows the various sensor fusion methods to be tested in a modular fashion, is also presented, along with the results from field-testing.

This paper proposes a novel face detection method to be used for practical human interactive mobile robot. Towards future aging society, there are much expectation and social demand for a mobile robot that is possible to interactively collaborate and support human. Typical examples are pet robot and service robot. Guard robot is another example. Face detection and recognition are very crucial for such applications. However, in the real situation, it is not easy to realize the robust detecting function because position, size, and brightness of face image are much changeable. The proposed method solves these problems by combining correlation-based pattern matching, histogram equalization, skin color extraction, and multiple scale images generation. The authors have implemented a prototype system based upon the proposed method and conducted some experiments using the system. Experimental results support effectiveness of the proposed idea.

The primary data used in ground-based, global path planning for NASA's Planetary Rovers are stereo images down-linked from the rover and range data derived from those images. The range data are often incomplete: the sensors are inherently noisy and sections of the landscape are blocked. This missing data complicates the path planning process and necessitates the help of human experts. We present the Rover Obstacle Visualizer and Navigability Evaluator (ROVANE), which assists these human experts and allows non-experts to plan missions without expert help. ROVANE generates a hazard map identifying slow, impassable, or dangerous regions with varying degrees of certainty. This map is used to create possible paths, which are assigned variable costs based on possible hazards. A hazard visualization is also produced, allowing the user to visually identify hazards and understand the system's path selection. As target locations are entered by the user, the system finds appropriate paths using a variation of the A* algorithm. A found path can be further modified by the user and output in a format suitable for commanding an actual rover. The system is capable of stand-alone operation, but is designed to be integrated into the Jet Propulsion Laboratory’s Web Interface for Telescience.

The goal of this research is to quickly create a compact, accurate representation of the path of a robot. Others have developed techniques for building 3D maps, while this work concentrates on building a descriptive image of where the robot has been. This technique which we call "tubular mosaics," provides an orthographic-like view of the hallways the robot has traversed. An omnidirectional camera produces images that can contain a hemispherical eld of view. Mosaics are built from wedges extracted from each frame. We demonstrate the technique using views perpendicular to the camera motion resulting in the upper half of the tubular mosaic (i.e. two walls of a corridor, including the ceiling.) Each wedge is unwarped into a rectangular strip. Strips from consecutive frames of video are matched and incorporated into the mosaics. Others have created orthographic strip mosaics of areas using a camera looking down at an angle from an airplane. Our work is a generalization of this to multiple directions. Three techniques of varying complexity and accuracy are presented. Results will be useful for robot navigation. Other applications include texture mapping and the creation of a compact representation of the video stream captured by the robot.

A coverage path planning algorithm produces a path that directs a robot (or its detector range, occupied volume, etc.) to sweep out a target volume. The goal of this work is to direct an agent to cover (fill) an unknown space using different coverage patterns. We achieve coverage by dividing the target region into sub-regions, termed cells, such that coverage in each cell can be accomplished by basic maneuvers, such as simple back-and-forth motions. This approach to coverage employs a representation of the free space called an exact cellular decomposition, which already has been widely used for path planning between two points. In this paper, we define exact cellular decompositions where critical points of Morse functions indicate the locations of cell boundaries. Morse functions are those whose critical points are non-degenerate. Different Morse functions induce different coverage patterns. Between critical points, the structure of a Morse function is effectively the same, so simple control laws to achieve tasks, such as coverage, are feasible within each cell. This paper addresses the issue of how to cover a space with varying patterns, but it also suggests a common framework for many conventional path planners. A companion paper describes the sensor based implementation of this approach.

A new approach is presented that aids in the specification of behaviors in a reactive architecture by establishing the value of categorizing motor schemas in four distinct classes, based in part upon the perceptual features which drive them. One of these classes encompasses much of the prior work in guidance behaviors, and here we show in simulation how valuable such behaviors are in minimizing difficulties during the parameterization process of other behaviors. Guidelines for schema gains are derived from basic principles for each of the four classes, and notional concepts of GUI interfaces are provided that take advantage of the guidelines. Although some of these issues have been addressed in other ways, the approach described here focuses on computationally efficient mechanisms that work in the true spirit of reactive control and can complement learning methods and hybrid architectures in general.

Newborn animals go through a period of adaptation and learning before they reach their full sensory-motor capa- bilities. A similar development strategy could be advantageous for newly built autonomous robots. In this paper we describe how a robot equipped with a generic, adaptive sensory and motor system con gures itself and acquires target approach behaviour purely from the exposure to sensory stimulation. Selforganising feature maps (SOM) and reinforcement learning provide the necessary adaptive mechanisms. These mechanisms are generic in the sense that they contain minimal dependencies on the actual devices used for sensor and actuator implementation. Our results demonstrate the behaviour acquisition can be achieved in autonomous fashion, in real time on a real minirobot.

One way to improve the efficiency of a mine search, compared with a complete coverage algorithm, is to direct the search based on the spatial distribution of the minefield. The key for the success of this probabilistic approach is to efficiently extract the spatial distribution of the minefield during the process of the search. In our research, we assume that a minefield follows a regular pattern, which belongs to a family of known patterns. Likelihood and Bayesian approaches to the pattern extraction algorithm are developed to extract the underlying pattern of the minefield. Both algorithms perform well in their ability to catch the "actual" pattern. And both algorithms are efficient, therefore, online implement of the algorithm on a mobile robot is possible. Compared to the likelihood approach, the advantage of using a Bayesian approach is that this approach provides information about the uncertainty of the extracted "actual" pattern.

This paper describes a new 3D sensing method for getting geometrical parameters of planar surfaces and cylindrical objects in a scene. The sensor system consists of two or three TV cameras with point light sources being placed at the projection centers of the cameras. The light sources are switched on alternately, and images under each lighting are recorded with the cameras. Under a point light source, even on a planar surface, there is a distribution of luminance due to diffuse and specular reflections. Since the light is emitted from the projection center, the positions of the peak of the luminance distribution on each surface directly provide such geometrical information on the surface. Our approach is based on measuring the peak positions on the images. This method speeds up considerably the acquisition of geometrical information on an entire scene, because the geometrical information on target objects can be obtained without analyzing range maps. The equipment setup for emitting light from the projection center is very simple and the illuminators can be easily set up at ordinary TV cameras. The proposed technique would be useful for real-world robotic applications such as navigation of indoor mobile robots. The experimental results show that the method is adequate for such purposes.

This paper describes applications of and infrastructural technologies for telepresence capabilities within the context of the Information-Based Manufacturing Program at the National Institute of Standards and Technology (NIST). The Information- Based Manufacturing Program is supported by the National Advanced Manufacturing Testbed (NAMT), which enables collaborative research among scientists and engineers at NIST, remote industry locations, and government and university laboratories. The research environment ofthe NAMT operates as a collaboratory, or a "research center without walls," using computing and networking technology to improve the performance of manufacturing research and the delivery of NIST measurement services. Telepresence is one of the critical technical capabilities upon which the operation of the NAMT collaboratory is based.

The mass appeal of space exploration on an interactive, personal level, is readily demonstrated by the popular success of the Mars Pathfinder mission, and the availability of "live" pictures over the internet. Over a half billion accesses to the NASA sites were recorded during the first 30 days on Mars, with a peak of over 46 million accesses on July 8, 1997 alone. The proximity of the Lunar surface provides near realtime opportunities for telepresence relying only on previously demonstrated technologies. A stereoscopic-vision Virtual Reality approach provides the sense of "being there" without the difficulties of getting there, and back. Internet access provides Earth-bound users with a hierarchy of potential levels of interaction, from vehicle driver, to active viewer, to passive viewer, from the Moon. An access fee structure can make such a venture economically self supporting when conducted on a sufficiently large scale [1 ] . A base- line study is presented consisting of a small fleet of 10 vehicles, each with a compliment of 50 remote-directed stereoscopic camera heads. Technical issues such as the problems encountered in an on-going operation on the Lunar surface are discussed. Consequences of temperature extremes, high vacuum, solar radiation, micrometeorites, abrasive nature of Lunar dust and its wide particle size distribution and the ability to electrostatically attach to surfaces are considered as they affect vehicle design and operation. An optical data link is described.

Handling heavy and bulky loads in manufacturing settings is an ongoing problem in achieving flexible and re-configurable assembly operations. One current trend is toward using powered heavy lift assist devices to augment human decision making skills with the mechanical muscle necessary to move and position large work pieces. As advanced automation has made it possible to apply power to multiple axes of a lift assist device, a need has arisen to implement a more natural control interface for the operator. One method is to use sensors to measure the force applied by the operator as the motion command to the powered assist device. This approach allows an untrained operator to easily guide the work piece using natural hand motions. In this research, we explore the use of multiple powered axes in a lift assist device to enhance and increase the material handling capability of the human operator. Input forces from the operator are measured and translated into joint actuation commands using a micro-controller. The automated control system is used to augment the workspace of to include virtual walls and attraction locations that can guide the operator as the work piece is moved. Closed loop control issues arise from the three coupled, nonlinear systems meeting at the manipulator input handle--the human, the powered lift assist device, and the digital computer control application. We present analytic and experimental comparisons of the performance of this class of human strength enhancing devices. Stability of the various control approaches is considered from a theoretical standpoint and compared to experimental results. The combination of theory and experiment are used to provide boundaries to the allowable performance of these types of machines.

It is the most important that we perform safe and reliable teleoperation of space robots in the experiment/operation using real satellite. Therefore we have developed and used Audio Feedback System (AFS) that use some auditory information such as motorized sound and prerecorded voice converted from a part of telemetry information to reduce loads to operators of space robots of the Engineering Test Satellite VII(ETS-VII). We applied AFS to the experiment ofAntenna Assembling Mechanism (AAM) which is experiment module of the Communications Research Laboratory (CRL) on the ETS-Vil for testing the mechanism of assembling structures in space. Our purpose of this study is to assess effectiveness of AFS by using the Eye Mark Recorder (EMR) which records eye movements for the astronaut and two operators as telemetry monitor who observe telemetry data. In results of macro shape of 0-500 (deg/sec) velocity distribution graph, average fixation time, and average velocity of eye movement, we have got important points as followings: ( 1) the spectrum patterns of velocity of eye movement of specialist operator of controlling space robots like an astronaut are significantly differed from other two operators, (2) the spectrum patterns of velocity of eye movement of specialist operator and nonexpert operator are not affected ofAFS, (3) the spectrum patterns of velocity of eye movement of well-trained expert operator was similar to specialist operator with AFS and similar to nonexpert operators without AFS, (4) average fixation time of specialist operator was not effected by AFS, average fixation time of well-trained expert operator, nonexpert operator became long without AFS support.

Deploying robotic devices for Web-based teleoperation is currently the domain of academic researchers, commercial enterprises and a small group of dedicated hobbyists. Before more people can deploy their own devices, a number of difficult issues need to be addressed. In previous work, we presented a reusable framework for Web-based teleoperation of robotic devices and attempted to address some of these deployment issues. However, a number of problems still remain, including how to support communication with a mobile remote device. Any remotely operated device requires some form of communication channel to its operator. In many current systems, this communications channel is composed of a long-haul network connection as well as a short-range connection between the device and a base unit. For a mobile device, one traditional solution for this short-range connection has been tethered operation. Tethered operation imposes a number of undesirable limitations on mobile devices however, so many researchers have turned to wireless communication as an alternative. As both the monetary and power costs associated with wireless communications continue to decrease, it becomes more viable as a general solution to the problem of providing communication between a base unit and a mobile device. We present an overview of the current state of the art in commercially available wireless communications and its implications for Web-based teleoperations. We also discuss how the requirements and implementation of our existing reusable framework will be modi ed in response to the new requirements imposed by wireless communication.

This article presents a preliminary version of the "Museum of Pure Form" , a virtual reality system where the user can interact, through the senses of touch and sight, with digital models of 3D art forms and sculptures. Two "Museum of Pure Form" systems are currently being developed at PERCRO, Pisa, Italy. One of these systems will then be tested inside several museums and art galleries around Europe by people normally visiting such institutions. The second system will be tested inside a CAVE system provided by another European institution. The system functionalities as well as its overall structure are presented.

Remote driving is a difficult task. Not only do operators have problems perceiving and evaluating the remote environment, but they frequently make incorrect or sub-optimal control decisions. Thus, there is a need to develop alternative approaches which make remote driving easier and more productive. To address this need, we have developed three novel user interfaces: GestureDriver, HapticDriver and PdaDriver. In this paper, we present the motivation for and design of each interface. We also discuss research issues related to the use of gesture, haptics, and palm-size computers for remote driving. Finally, we describe lessons learned, potential applications and planned extensions for each interface.

The symbiosis between computer graphics and modern planning and control methodologies is the basis for the development of projective virtual reality based telepresence techniques at the IRF. Virtual world appear close-to-reality for the user because of the interaction modeling technique borrowed from the field of robotic research. These techniques provide an intuitively operable user~environment for a greater range of application.The underlying idea of projective virtual reality is to first let the user work in a virtual world modeled after the physical plant to control and supervice. A projective virtual reality system then automatically deduce the impact of the user”saction on the state of the virtual plant and in turn employs action planning method to generate the equivalent impact on the physical plant using robots or other means of automation.Thus the robots are projecting the user “s action from the virtual into the physical world. The the German /Japanese project GETEX (German ETS-VII Experiment),the IRF realized the telerobotic ground station for the free flying robot ERA on board the japanese satellite ETS-VII. During the mission in April 1999the Virtual Reality based command interface out to be an ideally suited platform for the intuitive commanding and supervision of the robotic in space.

We report on the design issues of the User Interface for the MINIMAN system, a micro-robot with a sub millimetre precision manipulator. We introduce the project and its specifications alongside the requirements for the User Interface. The aim is to implement a semi-autonomous, computer vision controlled micro-robot, with a powerful interface. Having the user on a different scale than the robot is equivalent in terms of the user interface to a telecontrol configuration even if the user and the robot are located in the same room. Such a configuration requires a user/robot control loop, integrated in the interface, which address the following issues: The perception by the user of the current situation so that he become aware of what is the state of the robot and of the environment, the presentation of the possible actions and the mean to initiate them so that the user can plan and schedule his actions, and finally a feedback to the user when the robot has performed an action which lets the user become aware of the resulting effects of his actions. We have studied different approaches to the design of the user interface resulting from the need for a user-friendly robot destined to an industrial environment. The approaches of a particular interest are: Graphical User Interface, web interface and Virtual Environment. Although partially implemented at this stage they are described with analysis.

In the approach presented in this research, a six DOF industrial manipulator is used as the master device to provide haptic feedback to the operator. In order develop effective constraints between the motion of the slave and master, a virtual manipulator concept is developed that couples the actual robotic kinematics with the constraints of the simulated slave manipulator. The position and velocity errors between the actual and virtual mechanisms are used to develop an optimal impedance controller that constrains the motion of the master in all directions that are orthogonal to the allowable motions of the slave. This approach allows the use of a conventional industrial manipulator as an effective haptic display device.

Intention tremor severely compromises everyday life tasks in patients suffering by Multiple Sclerosis or Parkinson's disease. The present paper reports the current achievements in the analysis of subjects' movements. The trajectories produced by patients affected by tremor are investigated. Indexing procedures for the evaluation of the tremor which is present in the trajectories are proposed. Such indexes should be sensitive to tremor while they are not to voluntary movements. The properties of the indexes have been verified both with analytical considerations and with numerical tests. An experimental setup has been realized for the analysis of tremor on patients affected by Multiple Sclerosis tremor. The data collected within this experiment have been compared with expected results. Furthermore the indexing procedure has been applied for the evaluation of a filtering system. The case of handwriting has been chosen as reference task for the evaluation of achieved results. The results on clinical tests will be presented. Finally the applicability of the index-based procedures to the control of filtering systems will be discussed. The control architecture in the case of force-feedback less interfaces as well as haptic interfaces will be presented and discussed.

TORUS (for Toys Operated Remotely for Understanding Science) is an Internet-based educational project aimed at exploiting toys to create interesting and challenging robotics demonstrations and problem scenarios for students. In this paper we describe the implementation and evaluation of the TORUS "Construction Site" which aims to demonstrate key features of remote teleoperation via a multi-user cooperative task scenario. In this scenario three vehicles (toys) are remotely operated by three separate users to collectively complete a simple cyclical task. The users have separate control for their respective devices and the choice of three camera views. We describe the setup of the site, the control interfaces and the overall architecture of the Construction Site scenario. We report on initial evaluation of the scenario, the enhancements already implemented and others that are planned. One of the key features of TORUS is the use of toys as the remote "robotic" devices. The use of toys removes the usual costs and the safety requirements normally associated with using real robot devices. We discuss the scope for the further development of this approach and its potential for supporting both introductory and advanced robotics and artificial intelligence education via the Internet.

At the Institute of Robotics Research in Germany (IRF), an excavator- and construction machine simulator based on latest virtual reality technology has been developed. The main issues of the realization so far have been the real-time capability, the close-to-reality presentation of the environment and the physically correct simulation of the process, i.e. the simulation of the flow of the material handled with e.g. the simulated excavator. In the next step, it is being envisaged to enhance the system in a way that it cannot only be used for training, but also to command and supervise large construction machines in real world applications by means of virtual reality and automatic action planning components. Experiences gained from the control of space robots being controlled by methods of “projective virtual reality” will be introduced into this application to allow one driver to remotely control several excavators e.g. in a mining environment. In the paper, we describe how the simulation of the excavator and its interaction with the handled material can be handled mathematically and we will explain the basic ideas of how to “project” action that were carried in the virtual word onto physical excavators by employing the methods of projective virtual reality.

In this paper we present a multi-modal web interface for autonomous mobile robots. The purpose of this interface is twofold. It serves as a tool for task supervision for the researcher and task specification for the end-user. The applications envisaged are typical service scenarios like remote inspection, transportation tasks or tour guiding. Instead of post-processing a huge amount of data gathered and stored during operation, it is very desirable for the developer to monitor specific internal variables of the target system in real-time. Sensory information on several levels of abstraction are visualized using state-of-the-art web technology yielding a plug-in-free interface which can be viewed with a standard browser. It provides multi-modal information in several representations: off- and on-board vision, laser and odometry. This tool proved to be indispensable in the developing phase of navigation algorithms for localization, map building, obstacle avoidance and path planning. Modern guidelines for ergonomic interface design like context-sensitive popup menus or clickable goal specification make the interface very intuitive for the end-user. Its practicability is extensively demonstrated in the 'Computer2000' exhibition event, where during 4 days the robot was remote controlled in a fully autonomous mode by visitors of the tradeshow using this interface.

This paper presents a novel architecture allowing a generic force-feedback device to be used by different software tools dedicated to teleoperation and mission planning. This architecture relies on a “force-server” program running between the real-time controller of the haptic device and a set of applications using it. Possible applications include mission ground control system interfaces, telemetry systems coming back from real robots, or external simulation programs. The force-server concept is based on a high-level description of the forces to be generated. This description consists of spatial constraints defined by their type (point, line, plane and mesh), position and orientation. A force profile (space and/or time dependent) is assigned to each constraint. This description enables the generation of complex fields of forces by combining basic constraints. The advantage of this method is that an application can send force updates by simply modifying several parameters of the constraint. Between two updates (from the application) the force-server is able to continously compute new forces corresponding to the actual position of the device handle. This approach enables the control loop of the force feedback device to easily run at 500Hz when the application may send updates only at 25Hz. This novel method widens the use of force-feedback devices by providing a common interface to the different applications, and allowing multiple clients to use the same haptic device. A testbed using a 6 DOF haptic device has been developed. The device generates forces coming simultaneously from three different sources: a Java interface to experiment various force profiles, a rover simulator, and a scientific visualization tool used during planetary missions.

The PumaPaint project is a web robot that allows users to create original artwork on the World Wide Web. The site allows control ofa PUMA 760 robot equipped with four paintbrushes, jars ofred, green, blue and yellow paint and white paper attached to a vertical easel. Users must download a java TM interface allowing interactive control ofthe robot. This interface contains two windows showing live camera views ofthe work site and various controlsfor connecting and disconnecting to the robot, viewing the task status and controlling the painting task. The site operatedfrom June 1998 until March, 2000. This paper discusses the author 's experiences in operating the site throughout its lfecycle and discusses future plans for the PumaPaint project.

This paper presents an autonomous control frame for the virtual human. It introduces the object-oriented paradigms to model the action universe, and the planning paradigm to define the decision-making mechanism. From object-oriented idea, the real-world and operational models are elaborated. The real-world model consists of the human skeleton objects and the entities composing its environment, while the operational one focused on the perception, the behavior and the motor of virtual human. From the artificial intelligence research, the planning paradigm is adopted to constitute a basis for the mapping from a given task to a sequence of atomic actions. The use ofthis frame provides extensive application perspective for autonomous virtual human research.

Remote teleoperation allows humans to extend their capabilities to environments too dangerous for biological organisms. The communication between the operator and the robot needs to be fast in order to accommodate quick interaction with the environment. However, certain remote environments, such as deep-sea and space operations, impose restrictions on the communication link which limits the available bandwidth. In addition, as the distance increases, the communication between the operator and robot is further handicapped by significant time delays. The difficulties of controlling a robotic system remotely with long time delays and limited communication bandwidth are investigated. Four subjects performed a manipulator positioning task, controlling a simulated seven-joint manipulator under various time delay and communication rates. Results confirm performance degradation for 1.5, 3, and 6-second round trip delays. Compared to no delay, the time to complete the task increased between 160% and 480%. Limiting the communication bandwidth from 20 Hz also degraded performance, but not as severely as time delay. Only a 36% increase in completion time occurred when the bandwidth was reduced by a factor of eight to 2.5 Hz. A novel predictive display was introduced which dramatically decreased completion times to levels similar to no delay being present.

Designing mobile robots means more precisely, designing reactive autonomous systems. We de ne a reactive system as a system able to respond to a huge pannel of situations of which it does not have the schedule. In other words, for instance, the robot does not know when a particular situation may occur, and overall, what it would be doing at this time, and what would be its internal state. This kind of robot must be able to make a decision and to act even if it does not have all the contextual information. We do think that the prototyping phase is critical. Good and clever choices have to be made. Indeed, we may not easily upgrade such robots, and most of all, when the robot is autonomous, any change in both the software and the physical "body" is going to be very difficult, if not impossible. Thus, a great e ort has to be made when prototyping the robot. Furthermore, if your programming model is not expressive enough, you may experience a great deal of difficulty adding all the features you want, in order to give your robot reactiveness and decision making autonomy. For the last two years we have been experimenting, and teaching to our students, the Active Object Oriented Programming for such a purpose. To do so, we use a computer language named oRis featuring object and active object oriented programming, but also parallel and dynamic code, (the code can be changed during its own execution). This last point has been made possible because oRis is fully interpreted. However oRis may call fully compiled code, but also Prolog and Java code. An oRis program may be distributed on several computers using TCP/IP network connections. The main issue in this paper is to show how active objet oriented programming, as a modern extension of object oriented programming, may help us in designing autonomous mobile robots. Based on a fully parallel software programming, an active object code allows us to give many features to a robot, and to easily solve task con icts. This kind of programming is being studied by the French Army, and is taught in the French Army officers Academy. Active object oriented programming is also very useful in matters of software engineering. Indeed, inside the code, the separation between the logical parts is explicit and plain. This allows the designer to take only the robot's logical software part, regardless of the software testing environment, and to put it on the physical robot. And even among the logical parts of the robot software, the separation is quite huge, which is a good thing in terms of code engineering, upgrading and reusing.