`
`n ~~~ llIBIBIBrrr~unIkunIIDID lllIDftCSIflIDIDftit@lIDIDll W@IfIk~Thl@, @lID
`A~WIDlID©®~ m.@IID@ftit~
`
`- Can robots contribute to preventing environmental deterioration? -
`
`November 8-9, 1993
`AIST Tsukuba Research Center
`Tsukuba, Japan
`
`ENGINEFR!NG SOCIETIES LIBRARY
`
`AUG 8 - 1994
`
`LIBRARY
`
`1937
`
`IEEE
`
`Cosponsored by:
`Mechanical Engineering Laboratory (MEL)
`IEEE Industrial Electronics Society
`Robotics Society of Japan (RSJ)
`Society of Instrument and Control Engineers (SICE)
`
`Technically cosponsored by:
`IEEE Robotics and Automation Society
`IEEE Neural Network Council
`Japan Society of Mechanical Engineers (JSME)
`Japan Society of Precision Engineering OSPE)
`
`IPR2013-00419 - Ex. 1010
`Toyota Motor Corp., Petitioner
`
`1
`
`
`
`1993 IEEE/Tsukuba International Workshop on Advanced Robotics
`- Can robots contribute to preventing environmental deterioration? -
`
`Copyright and Reprint Permission: Abstracting is permitted with credit
`to the source. Libraries are permitted to photocopy beyond the limits of
`U.S. copyright law for private use of patrons those articles in this volume
`that carry a code at the bottom of the first page, provided the per-copy
`fee indicated in the code is paid through Copyright Clearance Center, 27
`Congress Street, Salem, MA 01970.
`Instructors are permitted
`to
`photocopy isolated articles for non-commercial classroom use without
`fee. For other copying, reprint or republication permission, write to IEEE
`Copyrights Manager, IEEE Service Center, 445 Hoes Lane, P.O. Box 1331,
`Piscataway, NJ 08855-1331. All rights reserved. Copyright © 1993
`by the Institute of Electrical and Electronics Engineers, Inc.
`
`Robotics Society of Japan (RSJ) and The Society of Instrument and
`Control Engineers (SICE) reserve the right to make limited distribution of
`the proceedings during and after meeting.
`
`IEEE Catalog Number:
`
`93TH0589-2
`
`ISBN:
`
`0-7803-1441-7
`0-7803-1442-5
`
`Softbound Edition
`Microfiche Edition
`
`Library of Congress:
`
`93-80011
`
`2
`
`
`
`Preface
`
`Welcome to the 1993 IEEEffsukuba International Workshop on Advanced Robotics. The
`workshop is subtitled "Can robots contribute to preventing environmental deterioration?" The
`word 'environment' has been used by robotics researchers for many years. It usually means a
`model space in which robots are supposed to work. In this workshop the word 'environment'
`has a different meaning. It no longer means a model space but the real space in which we are
`bound to live: the earth.
`
`We are now seriously aware that the resources of the earth are limited and we have to use them
`sparingly. We have to reduce pollutions by promoting recycling and by reducing refuse and
`waste. We have to maintain natural environments such as forests and seashores. We hope
`robots can help us to achieve this just like they do in manufacturing products.
`
`If we, the robotics researchers, want robots to be more useful, why don't we try to use robots to
`solve some of the environmental problems? This is the motivation of this workshop. Perhaps
`this is the first meeting in the world about robotics for the environment. We will call this
`'environmental robotics.'
`
`Robots are romantic machines and robot researchers are romantic people. Some of them may
`not like environmental robotics because environmental problems are not romantic, but the others
`may like it because the solution of environmental problems is romantic for it helps to create an
`utopia.
`
`In order to create environmental robots, we have to clear two steps. The first step is to make
`robots to perform certain necessary tasks. If they cannot perform these tasks, there will be no
`such robots. The other step is to justify the existence of such robots. They must save more than
`they consume. Should they be justified in terms of costs? Maybe, but costs depend on policies.
`Should they be justified in terms of natural resources? Absolutely. We will be better off without
`environmental robots if they consume more resources than they save. This is a severe demand
`for the robotics researchers.
`
`We do not have to be serious all the time talking about this problem. We can just enjoy the
`presentations, discussions, and our friendships in this workshop.
`
`Finally, I would like to thank all the participants of the workshop for coming all the way to
`Tsukuba from around the world. I am also grateful to all the workshop executives for their
`efforts to organize the workshop. Special thanks are due to the Foundation for Promotion of
`Advanced Automation Technology and the Electro-Mechanic Technology Advancing Foundation
`for their financial supports.
`
`Kazuo Tani
`General Chair
`
`LINDA Hf\LL L\ RARY
`
`3
`
`
`
`Workshop Executives
`
`Honorary Chair:
`Dr. Hisayoshi Sato, Ex-Director-General, MEL
`General Chair:
`Dr. Kazuo Tani, Mechanism Division, MEL
`Programming Committee:
`Dr. Kazuo Tanie, Biorobotics Division, MEL
`Dr. Kiyoshi Komoriya, Cybernetics Division, MEL
`Dr. Tatsuo Arai, Autonomous Machinery Division, MEL
`Dr. Kunikatsu Takase, Intelligence Systems Division, Electrotechnical
`Laboratory
`Dr. Mitsuo Wada, Human Informatics Department, National Institute of
`Bioscience and Human Technology
`Prof. Tamio Arai, University of Tokyo
`Prof. Shin'ichi Yuta, University of Tsukuba
`Mr. Yukiyoshi Hatori, Environmental & Safety Engineering Department,
`Nissan Motor Co. Ltd.
`Prof. P. Dario, Scuola Superiore S. Anna, Pisa
`Prof. A. Halme, Helsinki University of Technology
`Prof. R. Schraft, Head, Robotics Group, IP A
`Dr. G. Giralt, Robotics and AI Group Head, LAAS
`Advisory Committee Chair:
`Prof. Fumio Harashima, Director-General, Institute of Industrial Science,
`University of Tokyo
`Advisory Committee:
`Dr. Ken-ichi MatsUDo, Director-General, MEL
`Dr. Taketoshi Nozaki, Robotics Department, MEL
`Dr. Hideo Inoue, Manufacturing Systems Department, MEL
`Dr. Masao Kubota, Foundation for Promotion of Advanced Automation
`Technology
`Dr. Masakazu Ejiri, Mechanical Engineering Research Laboratory,
`Hitachi Ltd.
`Dr. Tsuneji Yada, Tsukuba R&D Laboratory, OMRON Corporation
`Mr. Hirotaka Miura, Tsukuba Research Laboratory, Yaskawa Electric
`Corporation
`Prof. Toshio Fukuda, Nagoya University
`Prof. T. J. Tarn, Washington University
`Prof. Robert Marks, University of Washington
`
`This workshop is supported by:
`
`Foundation for Promotion of Advanced Automation Technology (P AA T)
`Electro-Mechanic Technology Advancing Foundation
`
`4
`
`
`
`Table of Contents
`
`Global Environmental Problem and its Implication to Technology
`S. Nishioka (National Institute for Environmental Studies, Japan)
`Concept of Ecofactory
`M. Hattori, H. Inoue (MEL, Japan)
`Recycling on Network: an information-control architecture for ecologically-conscious industry
`K. Kamejima, M. Ejiri (Hitachi, Ltd., Japan)
`New Robot Applications in Production and Service
`R.D. Schraft, E. Degenhart, M. Hagele, M. Kahmeyer (IPA, Germany)
`Conceptual Design of Disassembly Automation System for Automated Manufacturing with Ecological
`Recycling
`T. Shibata, K. Tanie (MEL, Japan)
`The Concept of Robot Society and Its Utilization
`A. Halme, P. Jakubik, T. SchOnberg, M. Vainio (Helsinki University of Technology, Finland)
`An Experimental Robot System for Investigating Disassembly Problems
`P. Dario, M. Rucci, C. Guadagnini, C. Laschi (Scuola Superiore S. Anna, Pisa, Italy)
`Intelligent Robotic Technology for Environment Conscious Reusable Manufacturing
`T. Fukuda, K. Shimojima (Nagoya University, Japan)
`Motion Planning for Robotic Spray Cleaning with Environmentally Safe Solvents
`Y.K. Hwang, L. Meirans, W.D. Drotning (Sandia National Laboratories, USA)
`Recycling of Printed Wiring Board Waste
`S. Yokoyama, M. Iji (NEC Corporation, Japan)
`Vision System for Part Disassembly Using a High-Speed Range Sensor
`T. Arai, K. Umeda (University of Tokyo, Japan)
`An Image Recognition Method for Rusty and Damaged Car Parts After Road Traffic Accident
`K.H.L. Ho (University of Bristol, UK), K. Yamaba (MEL, Japan)
`Seashore Robot for Environmental Protection and Inspection
`T. Nakamura (Mie University, Japan), T. Tomioka (Suzuka College of Technology, Japan)
`Synthesis of Parallel Manipulators Using Lie-Groups: Y -STAR and H-Robot
`F. Sparacino (politecnico di Milano and Ecole Centrale Paris, Italy),
`J .M. Herve (Ecole Centrale Paris, France)
`Integrated Limb Mechanism of Manipulation and Locomotion for Dismantling Robot
`- Basic concept for control and mechanism -
`N. Koyachi, T. Arai, H. Adachi (MEL, Japan), Y. Itoh (Nisshimbo Co. Ltd., Japan)
`The Concept of Model Free Robotics for Robots to Act in Uncertain Environments
`K. Tani, K. Ikeda, T. Yano, S. Kajita, O. Matsumoto (MEL, Japan)
`Dynamics and Control of Aerial Mobile Legs
`T. Tsujimura, T. Manabe, T. Yabuta (NTT, Japan)
`ProLab 2 : a driving assistance system
`M. Rombaut (Universite de Tecbnologie de Compiege, France)
`Future Use of Robotics in Forestry
`C. Asplund (Swedish University of Agricultural Science, Sweden),
`A. Fukuda (Forestry and Forest Products Research Institute, Japan)
`Leg-Wheel Robot: A Futuristic Mobile Robot Platform for Forestry Industry
`Nakano E., Nagasaki S. (Tohoku University, Japan)
`Some Considerations on Robotics for Environmental Friendliness
`F.G. Pin (Oak Ridge National Laboratory, USA)
`
`1
`
`3
`
`9
`
`15
`
`25
`
`29
`
`37
`
`43
`
`49
`
`55
`
`59
`
`65
`
`69
`
`75
`
`81
`
`85
`
`91
`
`97
`
`103
`
`109
`
`Late
`
`5
`
`
`
`Proceedings of the 1993 IEEErrsukuba International Workshop on Advanced Robotics
`- Can robots contribute to preventing environmental deterioration? -
`Tsukuba. Japan November 8-9, 1993
`
`ProLab 2
`
`a driving assistance system
`
`M. Rornbaut *
`Heudiasyc URA CNRS il17 - UTe
`Centre de recherches de Royallieu, BP 649, 60206 Cornpiegne Cedex, France
`
`Abstract-~This paper deals with a driving assis(cid:173)
`tance system developed by the French Pro Art group
`of Prometheus. The system gives advices about risks
`and possibilities when the driver executes a maneu(cid:173)
`ver. It is embedded in a normal car (605 Peugeot of
`PSA). It is composed by a perception system and a
`decision one. This system diagnoses the situation and
`determines the potential or real risks. The resulting
`messages are sent to the driver via a "driver/vehicle"
`interface. The perception system determines the state
`of our vehicle, the state of the static environment
`(road, lines, ... ) ,and the state of the dynamical en(cid:173)
`vironment (moving obstacles, ... ). The global system
`works out in the standard driving situations in motor(cid:173)
`way, two carriage ways and in a crossroad in a town.
`
`1.1 The global functionalities
`
`ProLab 2 is a driving assistance system. Its goal is te, as(cid:173)
`sist the driver during a given maneuver, and to inform him
`about the potentiel or real risks involved. We assume that
`the others vehicles are the standard ones, with no specific
`equipment. The infrastl'llcure is also standard excepted
`active beacons that can informs the system about the
`static infrastructure of the road like the typ~~ of crossroad
`(the priority, the topology, ... ). None information is given
`about the moving obstacles or vehicles. The embedded
`system must understand the situation, foresee the driver
`behavior and the other vehicles' one, determine the real
`risks (over speed limit) or the potential ones ( dangerous
`overt.ake ). Then, it informs the driver on concequence.
`
`1
`
`Introduction
`
`The security on the road is a very important problem
`in the industrialised countries because of the deaths and
`injured, but also of the financial costs of accidents. In Eu(cid:173)
`rope, the "Prometheus" program has been created to find
`solutions for this problem. These solutions are studied
`for different types of systems. Ones improve the security
`inside the vehicle like ABS system but also systems to
`avoid slipping on ice. Others ones create a communica(cid:173)
`tion system between vehicles or between the vehicle and
`the road infrastructure. The subgroup" ProArt France" is
`currently developping a driving assistance system to help
`the driver in his maneuvers. The research group is com(cid:173)
`posed of nine research teams in France and two french cars
`compagnies (Renault and PSA). A firts demonstrator,
`ProLab1, has been developped on a R21 of Renault and
`presented at the Board Members' Meeting (BMM'91) at
`Torino in 1991. ProLab2 is the continuation of ProLabl,
`with more perceptive systems, more situations, dealt with
`more teams. It is developped on a 605 Peugeot of PSA.
`This demonstrator will be presented at the BMM'94 at
`Morte Fontaine (France).
`
`"e-mail rombaut@hds.ull.iv-compiegne.fr
`
`1.2 The global structure
`
`The system is decomposed in two main parts which per(cid:173)
`form the evaluation of the situation (perception part) and
`the interpretation of the sit.uation ( copilot part). To be
`able to interpret the sit.uation, it is important to get a
`"good image" of it. Then, some perceptioll systems in
`the demonstrator evaluate the situation of the vehicle (
`speed, acceleration, ... ) , the static environment situation
`(road, lines, ... ), and the dynamic environment situation
`(position of the others vehicles, ... ).
`The provided information is fused (temporal multi(cid:173)
`sensor fusion) and studied to maintain the colwrenc,' of
`the situation, as, for instance, in case of vehicle',; disap(cid:173)
`pearance in a blind space. The goal is to give to the
`copilot, the better image of the real situation. A dynami(cid:173)
`cal image of our vehicle (potential acceleration, ... ) is also
`built.
`The copilot analyses this situation and when a risk is
`detected, a message is transmitted to the driver. A rep(cid:173)
`resentation of the architecture is given figure 1.
`
`2 The copilot
`
`The copilot is based on the current analysis situation ,
`and its possible evolution. The risks can be deduced,
`and t.he messages are sent to inform the driver. It also
`allows to foresee the future situations and to calculate
`
`0-7803-1441-7/93/$3.00 © 1993 IEEE
`
`~)7
`
`6
`
`
`
`Dynamic data
`manager
`
`Copilot
`
`Driver/vehic1e
`inteface
`
`Figure 1: The global architecture
`
`the realisability or the risks of a maneuver. The global
`architecture of the copilot is presented at figure 2.
`
`Figure 2: The copilot architecture
`
`2.1 The situation diagnosis
`
`The goal of this module is to determine the global situ(cid:173)
`ation according to the information provided by the per(cid:173)
`ception modules. It is constituted by several real time
`expert systems, named SUPER [1]. The rules are hierar(cid:173)
`chicaly decomposed in several packages activated accord(cid:173)
`ing to the situation. This method allows to get speeder
`conclusions. The system accepts interrutions, and main(cid:173)
`tains the reasonning coherence. This mechanism allows to
`treat in priority the important events which disturb the
`evolution.
`According to the situation, we deduce the actions to
`work out, for exemple to send a message, or to change
`the sensors working mode.
`
`All along the time, the system evaluates the irnliledi(cid:173)
`ated risks of the situation. For any situation, t h<, p'ltell(cid:173)
`tial risks are continuously tested. For instance. (hirillg
`a pedestrian passage approach, a pesdestriall presellce is
`tested. In such a case, the speed should of course 1.(' re(cid:173)
`duced.
`
`2.2 Perception control
`
`Environment perception, especially artificial vision,
`IS
`highly time consumming, hence, it is not possible to COJl(cid:173)
`tinuollsly activate all the perception systems. Accord(cid:173)
`ing to the situation, some data sets are more important.
`Then, the perception system must be controlled to focus
`on them for a moment. The different data acquisitions
`are the follow:
`
`• continuous observation : the data evolution IS ob(cid:173)
`served up to be stopped.
`
`• one-off observation : the data is obs('fved only olle
`time
`
`• event detection: an event can change the n'asonning
`so it is important to detect it (expl: exceeding speed
`limit). The per.ception system interrupts the copilot.
`
`2.3 Help for manoeuver
`
`This system can help the driver during or before the ma(cid:173)
`neuver excution. It estimates if it is possible or danger(cid:173)
`ous. This system is decomposed in three parts: a plallning
`system which calculates the optimal realisation of the ma(cid:173)
`noeuver, an execution monitoring which checks the actual
`execution to the planned one [2], and a danger analysis
`module which verifies the security distances.
`If no possible solution is find, the planning system in(cid:173)
`forms the copilot of the risks. Otherwise, if the difference
`between the actual and planned realisation are too dif(cid:173)
`ferent, the execution monitoring system asks for a new
`calculation with new actual data to the planner.
`Those analysis are done at the situation diagnosis mod(cid:173)
`ule's demand for the planning step, and periodicall) with
`a 1 second sampling for the execution monitorin~. In
`addition, the danger analysis module checks almosl COIl(cid:173)
`tinuously (sampling period at 200ms) the safdy distances
`with respect to the surrounding obstacles independantly
`of the plan currently executed.
`
`2.4 pilot message sender
`
`For all tlH' modules, the pilot message sender recewd dif(cid:173)
`ferent messages which can be futher, redundant or even
`inconsistant. For instance, one can ask for acceleration
`because of an overtakE', and another one ask, for a decel(cid:173)
`eration because of the speed limit.
`
`7
`
`
`
`The system must choose between these messages ac(cid:173)
`cording to the situation of the vehicle and send them to
`the driver/vehicle interface.
`
`• 2 stereo linear cameras placed beside the lights to
`detect front and lateral obstacle slike pedestrians and
`bicycles
`
`3 The perception system
`
`• 2 lateral CCD cameras to detect lateral vehicles in a
`crossroad
`
`The copilot needs information to understand and foresee
`the situation. This information must be the as complete
`and accurate as possible. This specification induces the
`use of a large number of sensors of various types and tech(cid:173)
`nologies. Neverless, the quantity of devices involved in the
`perception system is limited because of the space, the en(cid:173)
`ergy, the computing power, and the financial constraints.
`
`3.1 The sensors
`The environment of the vehicle is the same as the one of
`our personal car. There is no special equipment in the
`infrastructure as well as the other vehicles or obstacles on
`the road. So, the system must be able to perceive as well
`as the actual driver.
`The first type of data perceived by the driver concerns
`the sensations when its vehicle is moving, like accelera-
`tion, braking, turning, the rate of engine rotation. Some
`proprioceptive sensors are used in the vehicle like lateral
`and longitudinal acceleration sensors or brake state sen-
`sor. These sensors give information about vehicle internal
`state.
`The driver displaces it vehicle on the road according to
`the driving rules. He must then know what type of road
`he is driving on, what line mark is the border of the lane
`and he should always be able to localise its vehicle in the
`lane. A CCD camera has been placed beside the rear view
`mirror like on figure 3 to get this information.
`
`Figure 3: The front camera
`
`The driver must also take into account the surround(cid:173)
`ing environment composed of the other vehicles, and the
`different obstacles like pedestrians, bicycles, and animals.
`To estimate the situation (position and velocity) of the
`obstacles, we use several sensors placed in our vehicle:
`
`• a 3D sensor composed by a laser telemeter associated
`to a CCD camera placed in the bumper to detect
`front vehicles
`
`\l9
`
`• a rear CCD camera to detect rear vehicles
`
`These different sensors are presented in figure 4.
`
`Figure 4: TIlt' obstacles sensors
`
`All the information is refered according to the vehicle
`reference.
`
`3.2 Static environment
`
`The static environment information is given by the front
`and rear CCD cameras. The software modules of the front
`one give:
`
`• the number and the location of the lanes and the type
`of lines
`
`• the vehicle's position and orientation according to the
`lane
`
`• the special horizontal signs like stop bend, arrows,
`peclestrian crossing
`
`The method consists of an outline extraction, and an
`linear approximation to detect the lines.
`The software modules of the rear camera give:
`
`• the structure of the road
`
`• the lane positioning
`
`• the type of the lines
`
`The image is segmented in several inteft~sting areas.
`Then special marks are searched. The software has been
`implemented on special morphologic real time computer
`based on all ASIC chip (see [3]).
`
`8
`
`
`
`3.3 Dynamic environment
`
`4.1 The temporal multisensor fusion
`
`These perception systems give information about moving
`obstacles independently of our vehicle motion. Several
`systems are used.
`
`3.3.1 TelemeterfCCD camera sensor
`
`This sensor is composed of a laser telemeter associated to
`a CCD camera. It has two ways of working:
`
`• detection mode : all the front space is scanned and
`the relative position of the obstacles is given
`
`• focused mode: the laser beam is focused on a specific
`obstacle and its relative position and velocity is given
`
`The method consists of an image partitioning in differ(cid:173)
`ent depth plans corresponding to the different obstacles.
`The software algorithms are implemented in a transputer
`based system named Transvision [4].
`
`3.3.2 Stereo linear cameras
`
`This perception system allows to detect all sorts of obsta(cid:173)
`cles in front of our vehicle. It is composed of two linear
`cameras . The algorithms give the relative position and
`velocity of the front obstacles. The method is based on
`an edge detection on the two images to determine the
`interest points, and to match them.
`
`3.3.3 Lateral CCD cameras
`
`The lateral cameras allow to detect the vehicles on the
`transversal roads in a crossroad. The algorithmes are
`based on a spatio-temporal segmentation, then an inter(cid:173)
`pretation of the dynamical areas. The method is pre(cid:173)
`sented in [5].
`
`3.3.4 Rear camera
`
`This camera is used to detect the rear obstacles. The
`algorithms allow to give position of these vehicles. They
`are the same that ones used in §3.2. These algorithms are
`also used for images from other cameras.
`
`4 The dynamic data manager
`
`The sensors work in asynchronous maner with different
`frequencies. The given information is sometime redun(cid:173)
`dant, sometime further. Some data can be unpresent be(cid:173)
`cause of the perception system. So, it is important to
`manage as well as possible the data to give to the copilot
`the best image of the actual situation. These modules are
`developped [8] by Heudiasyc (CNRS-UTC) and INRIA
`Sophia Antipolis.
`
`The system updates the image of the environment by a
`estimator / predictor filter when a new data is given by
`the perception system. According to the situation, a reli(cid:173)
`ability number is associated to each vehicle corresponding
`to its real presence. For instance, if a vehicle comes ill the
`blind zone, its number st.ays at the same value, because
`we cannot see it, but it really exists.
`With this filter, we can predict the future situation for
`a time sufficient for the copilot planning module presented
`in (§2.3).
`
`4.2 The dynamic module
`
`The calculated parameters represent the risks state of the
`vehicle like the security distance between the vehiclef>, the
`maximal speed in a curve or the wheels' slip. They depend
`on the dynamic state but also on the situation of the
`others vehicles. A 3 degrees of freedom dynamical model
`of the demonstrator has been elaborated for this purpose
`[6].
`
`4.2.1 Data recognition and transmission
`
`To understand the situation, it can be usefull to know
`the type of the obstacles (vehicle, truck, bicycle, pedes(cid:173)
`trian). According to their geometrical characteristics and
`behavior, its is possible to estimate their type.
`After that, the data must be sent to the copilot iII the
`manner to be easy used. For instance, the obstacleK are
`placed in interest zones like in figure 5. Depending of its
`zone, a vehicle must more or less important.
`
`Figure 5: The interest zones on highway
`
`4.2.2 The perception system controler
`
`According to the copilot needs, this module manages the
`sensory devices set hardware and software means, til!'
`hardware ones as well as the sofware ones. Getting a hight
`level information often needs the activity of several per(cid:173)
`ception modules. Sometime, it is possible to choose the
`modules to use, sometime, the needed modules art' not
`available, and sometime, the module must be shared be(cid:173)
`tween different activities. The goal of the perception sys(cid:173)
`tem controler is to choose the available perception mod(cid:173)
`ules to give the copilot the best possible information.
`This module also controls the activity or working modes
`of the different modules. That means it gives working
`
`--100
`
`9
`
`
`
`[2] Th. Fraichard and C. Laugier. Path- Velocity DecomposI.tion
`Revisited and Applied to Dynamic Trajectory Planning. icra,
`Atlanta, GA (USA), Mai 1993.
`
`[3] X.Yu, S.Beucher and M.Bilodeau. Road Tracking, Lane Seg(cid:173)
`mentation and Obstacle Recognition by Mathematica/ Mor(cid:173)
`ph%gy. In Proc Intelligent Vehicles '92 sYIllposium, Detroit,
`July 1992.
`
`[4] F.Collange, J.Alizon, J.Gallice , and L.Trassoudaine. A ,'am(cid:173)
`era Telemeter Multisensory System for obstacle Detection and
`Tracking. In Intelligent Vehicle Highway systems, 25th ISATA
`Silver Jubile, Florence(Italy), 1-5 June,1992.
`
`[5] R.Canals, J.P.Derutin, and F.Heitz
`SegmeJ,.tation spatio(cid:173)
`temporelle sur machine para/lele de vision TnANSVISJON.
`In 14e Colloque GRETSI, Juan Les Pins (France), Sept. 1993.
`[6J A.Alloum and M.Rombaut. A safety indicato'r sy<~tem for driv(cid:173)
`ing assistance. Road vehicle autoIllation, Bolton England, may
`1993.
`
`[7] P.Pleczon, and A.Kessaci. A Human-Machin, Inter/ac, for
`Driving Assistance. In Intelligent VelUcle Highway systems,
`25th ISATA Silver Jubile, Florence(Italy), 1-5 June,1992.
`
`[8] B.Eleter, and M.Rombaut. On-board real time driving sys(cid:173)
`In Road velUcle automation ROVA'93,
`tem architecture.
`Bolton(England), 24-25 May,1993.
`
`orders to the different modules. The working modes are
`continuolls mode, one time mode or event waiting mode.
`In this mode, the module works until the searching data
`appears. For instance, the question can be "give me the
`position of the next pedestrian crossing". The module is
`searching for it until it appears then calculates its position
`an stops.
`
`5 The copilot / vehicle interface
`
`This module transforms the information to be under(cid:173)
`standable by the driver. The system is designed to be
`fast undrstandable by the driver. It should be felt as a
`smart help rather than a constraint. A study has been
`made by [7] and has shown 3 help levels :
`
`• The alarm mode: the system informs the driver when
`the situation is or becomes dangerous,
`
`• the advice mode: the information is given when the
`driver asks for it. For instance, the question can be
`"is the overtake possible?",
`
`• the assistance mode: a type of driving can be pro(cid:173)
`posed to the driver like economical driving for exern(cid:173)
`pIe.
`
`The interface is realised with visual systems, but also
`with phonic ones.
`
`6 Conclusion
`
`This paper brievely presents the basic principles of the
`ProLab 2 demonstrator. This one is constituted by a
`large number of modules that works out all together and
`collaborate to give an efficient help to the driver during
`various driving maneuvers. The system can be used in
`a several real traffics conditions. The driver keeps the
`control of its vehicle, and can choose the type of help
`that he needs.
`The different modules are being implemented in the
`target system to be embedded in the real vehicle which
`is a 605 Peugeot of PSA corporation. The demonstra(cid:173)
`tor will be shown at the Board Members' Meeting at
`Morte Fontaine (France) on October 1994. ProLab 2 is
`the only demonstrator, with ProLabl (R21 Renault) to
`be realised by university laboratories group in Europe. In
`this demonstrator, the works of 9 different laboratories in
`France are implemented.
`
`References
`[1] N.LeFort , E.Piat and D.Ramamonjisoa. Toward a copilot
`Architecture Based on Embedded Real Time Expert System.
`In Proc. 1st Intelligent Autonomous Vehicle '93 SYIllposiuIll,
`Southampton, Great Britain, 1983.
`
`101
`
`10
`
`