 |
|
Publications, papers published in October 1999/communications publiées en octobre 1999
- 1
-
D. Alazard, A. Bucharles, G. Ferreres, J.-F. Magni, and S. Prudhomme.
An integrated methodology for flexible aircraft control design.
In AGARD RTO Structural Aspects of Flexible Aircraft Control,
Ottawa (Canada), 18 - 21 October 1999. RTO.
The paper aims to give an overview on recent research activities at
the Systems Control & Flight Dynamics department of ONERA in the field of
flexible aircraft control. A long term research program has been conducted
for several years, with funds from SPAE in Paris and with the technical
support of Aerospatiale Avions in Toulouse, France. Beyond the primary
objectives of reaching various specifications for simulataneous aircraft
flight and structural dynmaics control more fundamental questions have been
addressed concerning the implications of rigid structural dynamics coupling
for the selection of suitable control methodologies; These questions will be
developed in the paper, in two main chapters.
- 2
-
G. Hardier, C. Lambert, and S. Kummer.
The cogite system of the charles de gaulle aircraft carrier : an
innovating heel compensation device.
In 12th Ship Control systems Symposium, La Hague (Hollande), 19
- 21 October 1999.
To improve the seakeeping performances of the french nuclear aircraft
carrier Charles de Gaulle, with respect to its predecessors Clémenceau and
Foch, DCN has developed a platform motion control system, called SATRAP,
which reduces the ship motions while ensuring navigation capability. This
system involves a centralized computer which controls three subsystems: two
pairs of stabilizing fins, a set of rudders and a moving-mass system (athwart
ship) for heel compensation, the Cogite system. Originally conceived for
situations where the stabilizing fins would be otherwise saturated (steady
heel from wind or heavy weights displacements), Cogite has evolved into a
system which also improves the operational capability of the ship by limiting
the heel during sharp turns, permitting non-stop handling and preparation of
aircrafts. This paper presents the results of the studies which led to the
development of a performing mode of operation for SATRAP, where the helmsman
has full control of the rudders while the fins and Cogite system
automatically reduce the ship motions. From the operational requirements of
an aircraft carrier, it shows how the design stage has ended in defining this
innovating heel compensation device, and why the control laws structure was
mostly conditioned by the actuators peculiarities and desired performances.
The following points are discussed in this paper: operational requirements in
terms of performances, SATRAP system architecture and principles chosen for
the control laws (task allocation for the actuators, feedforward-feedback
controllers). Results obtained during the first sea tests, aboard the Charles
de Gaulle, are given to illustrate the system performances.
- 3
-
B. Escande and Y. Aurenche.
Trailing vortices and safety.
In CESA/AAAF Forum : Research for Safety in Civil Aviation,
Paris (France), 21 - 22 October 1999. AAAF.
The hazardous nature of wake vortices was initially observed with
aircraft on the landing approach when lining up at separation distances
considerably below those en-route. More recently, wake vortices were also
encountered en-route where vertical separation distances have been relaxed to
accommodate a growing traffic. Nowadays, the wake vortex issue has aroused
widespread interest, not only in the aviation community but also in the
general public. The first reason is today's saturated air traffic situation
during landing approaches at major airports in the world (Chicago, Dallas,
London-Heathrow, Frankfurt, ...) and on densely flown routes over the North
Atlantic, for example. The second reason concerns the potential hazards
created by the "Very Large Transport Aircraft" (VLTA), a future airplane
to enter airline service within the next decade. In both cases, solutions for
relaxed horizontal and vertical separations have to be found with no
compromise with respect to flight safety.
- 4
-
Mag. Barbier, N. Alquier, P. Fabiani, J.-L. Farges, C. Tessier, and A. Vincent.
Situation assessment, planning and safety of airport ground
movements.
In CESA/AAAF Forum : Research for Safety in Civil Aviation,
Paris (France), 21 - 22 October 1999. AAAF.
Aircraft ground movement safety is ensured by controllers who allow
mobiles to proceed depending on the global situation and intents of other
mobiles. In order to perform this task the controller relies on his own
experience. Nevertheless a total or partial automation of the task may be
considered. This study is related to the Advanced Surface Movement Guidance
and Control System (A-SMGCS) and addresses two main topics; the automatic
detection of runway incursions (situation assessment) and the planning of
movements on taxiways (decision). This paper presents topics that are
currently studied within ONERA's project "Aéroport du futur" by the
Control and Decision research unit.
- 1 - Situation
assessment: Situation assessment aims to go further than a mere numerical
tracking of mobiles (aircraft and vehicles) and to provide a tracking of
their activities . Therefore the situation is assessed in terms of on-going
plans and allows an early prediction of conflicts to be made . The modeling
of activities and plans (for instance aircraft departure) is performed thanks
to the formalism of colored Petri nets. Alternative activities and
uncertainty are taken into account within the plans. The tracking of the
activities is given by the marking of the Petri net when sensor or controller
information trigger the transitions. Topographical resources of a given
airport (runways, taxiway s) are modeled by another Petri net that describes
their spatial sequences. The interpretation of the first kind of net allows
the resources to be associated to each activity within the plan. Early
conflict detection is performed through temporal activity prediction and
associated potential use of resources. From the theoretical point of view
this amounts to compare the reachable markings of both kinds of Petri nets.
This method has been tested on actual data from Orly airport.
- 2 -
Decision: Depending on the result of the situation assessment movement orders
have to be planned on-line. A prediction model is used by an optimization
algorithm in order to provide periodically the orders to be sent to the
pilots. This optimization takes into account safety constraints; in
particular collisions must be avoided. The prediction model includes the
different types of orders, the anti-collision constraints and the temporal
evolution of the uncertainty associated to the position of each aircraft. The
optimization algorithm is a simulated annealing and at each sampling time the
first order is issued and the optimization is performed again using the
updated parameters of the assessed situation.
The method has
been implemented together with a simulation model and tested on Blagnac
airport data. First results shows the feasibility of the approach.
- 5
-
H. Fargier and V. Galvagnon.
Le problème du PERT flou dans les graphes série-parallèles.
In LFA '99 : Rencontres Francophones sur la logique floue et ses
applications, Valenciennes (France), 21 - 22 octobre 1999. Université de
Valenciennes.
Un problème de PERT est défini par un ensemble de tâches à exécuter
en respectant des contraintes de précédence, et en minimisant la date de fin
de la dernière tâche. Lorsque les durées des tâches sont mal connues et
modélisées par des ensembles flous, il n'est plus possible d'utiliser une
méthode de propagation arrière pour calculer les ensembles flous de valeurs
correspondant aux dates de début au plus tard et aux marges. Nous proposons
ici une méthode qui permet de résoudre correctement le problème du PERT flou,
y compris le calcul des dates de début au plus tard et marges, lorsque le
graphe de tâches est série-parallèle.
- 6
-
S. E. Salcudean, K. Hashtrudi-Zaad, S. Tafazoli, S. P. DiMaio, and C. Reboulet.
Bilateral matched-impedance teleoperation with application to
excavator control.
Control Systems Magazine, 19(5), October 1999.
This paper addresses issues of transparency and implementation of
dual hybrid teleoperation. A method for automatically adjusting the master
and slave impedance to match stiff and soft environments and to interpolate
between them is presented and evaluated using simulations. The application of
this techique to the force feedback control a a mini excavator is also
presented and discussed including supporting experimental results.
- 7
-
F. Boniol, J. Foisseau, and G. Bel.
De l' asynchrone vers le synchrone : modélisation et vérification de
systèmes embarqués distribués asynchrones.
In Journées Validation des Logiciels Réactifs VLR99, Toulouse,
1 - 1 October 1999. CNES.
A faire. Voir DTIM.
- 8
-
D. Farcy, O. Renier, and D. Tristrant.
Impact of new aerodynamic control concepts on flight qualities and
maneuverability of fighter aircrafts at high angles of attack.
In RTO : Flight in a hostile environment, Solomons Island,
Maryland, USA, 19 - 21 October 1999. RTO. Systems Cocepts and Integration
(SCI)Panel.
Apex blowing for the purpose of active control of the flowstream has
been investigated on both a 12% wing model and a 12% aircraft model, based
on a 60° sweep angle delta wing, single fin, slender forebody nose generic
fighter configuration, in the 4m-diameter vertical wind-tunnel of the ONERA
Lille. Forces, moments and skin pressures are measured over a [0-90°] angle
of attack range. Two independent blowing systems external to the model
provide mass flow rate from 0 to 9%. Optimization of the concept
configuration leads to blow downstream in front of respectively the 28% and
60% of leading-edge discontinuities for lift and roll enhancement. The
concept mainly provides efficiency in roll around the stall angles [25-40°],
a significant increase in lift at 40° and some more complex effects on the
pitching and yawing moments with opposed tendencies around 28°. Coupling
effects between the concept and ailerons or forebody strakes configurations
remains low. Blowing enhances static stability in roll and extends the
stability domain to larger angles of attack, limiting the unstability in yaw
between 30 and 40°. Blowing efficiency is maintained or even increased when
forebody strakes are deployed. Blowing and ailerons show comparable LCDP
characteristics in the high angles of attack range. Blowing mainly
demonstrates capabilities to globally extend the (a ,b) stability domain for
strakes-off configurations, and to extend the sideslip range of this domain
when strakes are fully deployed.
- 9
-
A. Gaillet.
Automatic obstacle avoidance by a mobile robot.
In 30th International Symposium on Industrial Robots. ISIR'99,
Tokyo (Japon), 27 - 29 October 1999.
ALAIN doit mettre copier le résumé.
- 10
-
A. Gaillet.
Automatic obstacle avoidance by a mobile robot.
In ICAS'2000 Proceedings, Tokyo (Japon), 25 - 29 October 1999.
The paper we want to present in the frame of the 30th International
Symposium on Robotics will deal with both the theoretical and experimental
studies we conduct on the subject of automatic obstacle avoidance by a mobile
robot. Let now fix the context of the obstacle avoidance in our study.
Basically the process of the obstacle avoidance, in our context, consists for
the vehicle, moving initially on a desired trajectory, to detect the moving
obstacle by an onboard sensor, to avoid it automatically by moving away from
its imposed trajectory and to come back on its initial track. The trajectory
followed by the vehicle has to be as close as possible in relation to the
obstacle : this deviation from the desired trajectory has to be minimal in
order to reduce the time lost during the process while satisfying at any time
the required constraints on the distance between the robot and the obstacle.
In a first phase we test, through simple numerical simulations, different
approaches based on potentials. We test different combinations of such
potentials, both repulsive and attractive ones and combinations of these two
types. For repulsive ones, different shapes of potentials were tested and in
each case an evaluation of the lateral deviation of the mobile robot during
the avoidance phase was made. In a second step, a numerical simulation of the
six wheeled robot was built in order to validate in a more realistic way the
selected solution and to test other best suited ones. Finally an automatic
avoidance control had been designed based on a prediction algorithm of the
position of the obstacle. To evaluate the real performances and robustness of
this control law, a model of the camera dedicated to the detection of the
moving obstacle has been added to the simulator and noise on the matrix of
the detector has been introduced. Finally this last control law has been
implemented on the onboard computer of the robot and tested in a real
environment on the site of the lab, firstly by simulating the displacement of
the obstacle and secondly by detecting and identifying the state of the
obstacle through processing of the image taken by a camera in front of the
robot. The paper will also present the results obtained during these tests
and a video tape on the experiments.
- 11
-
J.-L. Farges and J. L. Gallego.
Stochastic P-timed petri nets and their use for traffic queue
estimation.
In IEEE International Conference on Systems, Man and
Cybernetics, Tokyo (Japon), 12 - 15 October 1999. IEEE.
Introduction Stochastic Petri nets are usually used for performance
evaluation and the state estimation of discrete state processes is usually
performed using the formalism of Markov chains. However the description of
the Markov chain corresponding to a real life system is often too complex to
be performed. For that reason it is interesting to investigate the
possibility to use a type of stochastic Petri net for filtering problems. A
target application for such a possibility is the estimation of traffic queues
in urban networks. Indeed, for this problem, most of the approaches are based
on Markov chains (Bar:79), (Bet:87) and (Gal:97). Stochastic Petri Nets
Mathematical Preliminaries Two operators used to manipulate densities of
probability of discrete variables are defined. The first operator corresponds
to the extension of operators defined for the variables. The second operator,
named composition permits to compute the density of probability of a variable
 from, on one hand, the density of probability of a variable  and, on
the other hand, the density of probability of knowing  . Probabilist
Transition Petri Nets A probabilist transition is defined as a transition
where, instead of assigning to the transition a single list of output places,
several lists of output places are defined and a distribution with a
probability for each list is associated. Extended Probabilist Transition
Petri Nets The extension proposed here consists in associating to each place
a distribution of timings. This distribution is discrete and a single time
increment is defined for all the network. Firing Algorithm The purpose of the
firing algorithm is to compute a new distribution of markings resulting from
the firing of an enabled transition and an initial distribution of makings.
This algorithm is obtained by applying the composition operator to, on one
hand, the initial distribution of markings and, on the other hand, a simpler
firing algorithms which permits to describe the distribution of markings
resulting from a single initial marking. Modeling of Traffic Flow Model for a
Single Link A single link is modeled by a probabilist timed place between two
deterministic transitions. The tokens in the place represents the vehicles.
The distribution of timings in the place represents the platoon distribution
in free speed travel time for the link. If the link is controlled, another
place, with an initial mark of one token, is connected with two arcs to the
downstream transition. The distribution of timings in the place represents
the distribution of headways at saturation flow. Finally the transition is
enabled by a green signal. Model of a Network of Links Interconnections
between links are modeled by probabilistic transitions with a distribution
corresponding to the turning movement ratios. For a probabilistic transition
corresponding to the output of a given link there is a set of output places
with interlink timings. Each of those places is connected through a
deterministic transition to a specific place which gather the outputs of
several links going to the same next link. This last place is connected to
the input transition of the link. Filtering Principle Use of Stochastic Petri
Net for Correction and Prediction The result of the firing of some
transitions of the Petri net is observed. It corresponds to the physical
measurement of the actuation of loop sensors by vehicles. When the result is
true (i.e. the firing induces actual changes in the marking) that means that
the transition was enabled. Thus, using the Bayes' rule, all markings which
do not.
- 12
-
C. Reboulet and S. Durand-Leguay.
Optimal design of redundant parallel mechanism for endoscopic
surgery.
In proceeding of the 1999 IEEE/RSJ, volume 3, pages pp
1432-37, Kyongju (Corée), 17 - 21 October 1999. IEEE.
A new kinematic concept of a redundant wrist with parallel structure
is presented. This new device transforms the movement of four actuators
acting along parallel directions of space, in a motion with three rotations
around a fixed point. The properties of this new device are very interesting
because they allow the direct mechanical coupling between a master system and
a slave device. This mode of coupling allows to realize teleoperation;
moreover its thread-like morphology makes it well suited to the applications
in the surgical field. This new redundant morphology is presented as well as
a complete kinematic analysis which has shown that actuator redundancy not
only removes singularities but also increases workspace while improving
dexterity. Finally a dexterity measure and the optimal design of the
mechanism are presented.
|
