• 专利附录
  • 专利说明
  • 权利要求
  • 类似技术
  • 同族专利
  • 引用文献
  • 法律状态
  • 优先权
    • 专利摘要:
    The method comprises: - a preliminary phase wherein said vehicle is aligned to engage said target position; a first phase in which a trajectory (83) to be followed, referred to as reference, is generated as a function of the state of the vehicle and of the target position, said state being defined by the current position and the orientation of the vehicle; a second phase in which, said reference trajectory being cut into sections, it is predicted at the start of each section and before the vehicle initiates a movement if said reference trajectory can be tracked as a function of imposed space constraints and slips lateral and / or longitudinal estimates; in a third phase, if said trajectory can be followed, the steering angle of said wheels (81, 82) and the linear traction speed of the vehicle are controlled according to the state of the vehicle and the lateral sliding and / or longitudinals, to join the centers of said wheels on the reference path (83).
    公开号:FR3038279A1
    申请号:FR1501414
    申请日:2015-07-03
    公开日:2017-01-06
    发明作者:Eric Lucet；Alain Micaelli；Francois Xavier Russotto
    申请人:Commissariat a lEnergie Atomique CEA；Commissariat a lEnergie Atomique et aux Energies Alternatives CEA；
    IPC主号:
    专利说明:

    y'f et y'r étant respectivement la dérivée, par rapport à la distance curviligne, de l’erreur latérale avant yf et de l’erreur latérale arrière yr.
    On considère également le vecteur d’étal
    On peut montrer qu’il existe des matrices A, P et B conduisant au modèle linéarisé défini par les relations suivantes : y = Px (1)
    (2) y' étant la dérivée, par rapport à la distance curviligne de y, β'f et β'ν étant respectivement les dérivées, par rapport à la distance curviligne, de βί et / r, le vecteur d’entrée
    exprimant ainsi l’évolution des angles de braquage des roues avant et arrière en fonction du mouvement du véhicule.
    Les matrices A, P et B incorporent les équations du modèle cinématique illustré par la figure 8 tenant compte des entraxes df, dr.
    Pour l’observation de l’état du véhicule, on peut par exemple utiliser le vecteur :
    indiquant la position et l’orientation du véhicule par rapport au repère absolu (o), 6m étant l’angle du repère mobile (m), lié au véhicule, par rapport au repère (o) comme indiqué précédemment, xm et ym étant les coordonnées du repère mobile (m) dans le repère (o).
    Pour les observations de glissement, les vitesses et les positions des repères locaux (f) et (r) liés respectivement aux roues avant et arrière sont utilisés. La vitesse de rotation des roues sera plus particulièrement utilisée pour observer les glissements longitudinaux.
    Sur la trajectoire 83, on définit un nombre nsiep de tronçons, ou pas, curvilignes de longueur S. A partir du modèle linéarisé des relations (1) et (2), les matrices d’état du système continu sont Ac = AP-1 et Bc = B. En discrétisant un pas de longueur S le long de la trajectoire, les matrices du système deviennent alors Ad et Bd données par le système d’équations suivant (3) :
    U étant la matrice identité de dimension 4 x 4. Un calcul de puissance sur Ac montre que Ac3 est égal à 0. Pour des considérations de déplacement en marche avant et en marche arrière, les calculs sont faits pour S > 0 et S < 0.
    Les états prédits des vecteurs y sont calculés à partir de la connaissance de l’état courant yo et des matrices Ad et Bd.
    La prédiction sur les nsiep pas successifs conduit à la relation suivante (4) :
    (4) s’exprimant plus synthétiquement selon la relation suivante :
    (5)
    Les relations (4) et (5) montrent qu’à chaque pas de distance k, on définit une commande d’entrée Uk.
    Dans ce processus d’optimisation sous contrainte pour suivre la trajectoire 83, le critère à optimiser est par exemple une fonction quadratique du vecteur d’état prédit Y et de l’entrée de contrôle U. Ce critère, noté crit, s’exprime selon la relation suivante :
    (6) où Q et Jl sont les matrices de poids de dimension 4nstep x 4nstep et de dimension 2 nstep x 2 nstep respectivement pour l’état et le contrôle du véhicule. A partir des relations (5) et (6), on peut exprimer le critère selon la relation (7) suivante :
    (7) où le paramètre dépendant de yo est exprimé séparément.
    Les expressions des contraintes d’états pour un véhicule donné sont définies par la relation (8) ci-dessous. Plus particulièrement, selon une première approximation, le véhicule est modélisé comme un rectangle dont le côté avant et le côté arrière sont respectivement à une distance Df et de l’essieu avant le long de l’axe longitudinal 80. Le fait que ces côté avant et arrière restent dans une plage de tolérance Sgap autour de la trajectoire peut s’exprimer selon la relation suivante :
    (8)
    La plage de tolérance ôgap est de plus en plus contraignante en fonction des zones 71, 72, 73, 3, plus on s’approche de la position de parking visée.
    Cette contrainte peut encore s’écrire sous la forme matricielle suivante :
    (9)
    Etendue à tous les nstep états prédits et écrit comme une fonction des futures entrées de contrôle, cette expression devient :
    (10) où : • © est une matrice de nstep blocs disposés en diagonale, chaque bloc consistant une matrice D selon la relation (9). «
    L’obtention du vecteur d’entrée de contrôle U résulte alors de la solution au système suivant où le premier membre est relatif au critère à optimiser et le deuxième membre à la contrainte :
    (11)
    Avantageusement, la loi de commande est définie indépendamment du temps et de la vitesse.
    Pour passer à la loi de contrôle temps-réel des actionneurs, il faut ensuite faire intervenir le temps. Extrairepremier élément de la solution U, donne la loi de contrôle :
    (12)
    Vf étant la vitesse linéaire des roues avant.
    On obtient ainsi un système classique d’optimisation sous contrainte.
    La loi de contrôle définie par la relation (12) donne la dérivée des angles de braquage par rapport au temps, et indique donc au contrôleur les commandes de braquage à appliquer. Ces informations sont ensuite transmises via des interfaces adaptées pour impulser les mouvements aux roues.
    Avantageusement, la fonction d’observation est indépendante de la trajectoire à suivre et donc non impactée par les erreurs de suivi. Elle réalise trois types d’observations. Ces observations permettent la correction des dérives latérales et longitudinales par rapport à la trajectoire de référence, en temps réel.
    Un premier type d’observation effectue l’observation de l’état du véhicule défini notamment par sa position, son orientation et sa vitesse. Cet état peut être obtenu classiquement par un filtrage de Kalman étendu.
    Un deuxième type d’observation effectue l’observation des glissements longitudinaux du véhicule et plus particulièrement des roues avant et arrière. De façon pratique, cette observation peut être réalisée par l’observation du rayon apparent des roues, le rayon apparent étant calculé en fonction de la distance parcourue tenant compte des glissements longitudinaux. Le rayon apparent est plus grand ou plus petit que le rayon réel des roues, sauf en l’absence de glissement où il est égal. La comparaison entre le rayon apparent et le rayon réel permet donc d’observer le glissement longitudinal.
    Un troisième type d’observation effectue l’observation des glissements latéraux 8f, 8r illustrés par la figure 5.
    Pour ces différentes observations, des variables d’état proprioceptives, propres au véhicule, sont mesurées, notamment : - les vitesses de rotation moyennes oif et ωτ des roues avant et arrière ; - les angles de direction avant et arrière ay, ar illustrés par la figure 5 ; - les dérivées par rapport au temps ay, àr de ces angles, c’est-à-dire les vitesses angulaires de direction.
    Les variables extéroceptives mesurées sont notamment : - les composantes du vecteur
    indiquant la position et l’orientation du véhicule par rapport au repère absolu (o) ;
    La différence entre les mesures des variables proprioceptives, indiquant l’état théorique du véhicule, et les mesures des variables extéroceptives, tenant compte des glissements du véhicule, permet d’obtenir l’observation de l’état du véhicule, par exemple par un filtre de Kalman étendu. A partir de ces observations, les glissements longitudinaux et latéraux peuvent être estimés, comme l’illustre notamment l’estimateur de glissements 53 utilisant les données d’observation 52 dans l’architecture fonctionnelle illustrée par la figure 5. L’invention a été décrite pour un véhicule du type camion ou remorque porte containers. Elle s’applique pour tous types de véhicules nécessitant une conduite automatique vers une position cible, soumis à des glissements longitudinaux et/ou latéraux lors de leurs déplacements vers cette position cible.
    AUTOMATIC DRIVING METHOD FOR INSERTING AND EXTRACTING A VEHICLE IN A RECEPTION STATION,
    AND CONTROL DEVICE IMPLEMENTING SUCH A METHOD.
    The present invention relates to an automatic driving method for insertion and / or extraction of a vehicle in a docking station. It also relates to a control device implementing such a method. It applies in particular for the insertion or extraction of container trucks in docking stations dedicated to the installation and removal of containers.
    Intra-port traffic consists of moving containers between cargo loading / unloading points and temporary storage areas (linked to road and rail transport infrastructures). The loading / unloading phases are operated using mobile lifting devices (cranes) that will load / unload mobile platforms that will carry the transport between the various points of the port. The positioning accuracy of the mobile platform when parked is crucial to secure and accelerate the loading / unloading task.
    In fact, the parking operations of a platform or trailer, conveyed by means of a motorized cabin, are delicate and require a lot of time, even when they are carried out by experienced drivers. Accidents can also occur, damaging the materials contained in the containers but also to those involved in the surrounding maneuvers. To increase the productivity and safety of the port traffic, it is therefore necessary to speed up parking operations and significantly reduce the number of accidents. To reduce these accidents, it is necessary to load and unload the containers precisely, which requires precisely positioning the platforms in the parking spaces.
    To improve the accuracy and secure the loading and unloading of containers on platforms, there are parking stations equipped with side walls. The problem is then to quickly park a platform, trailer or long vehicle, in a parking area a few centimeters or even centimeter. Assistance for parking or automatic parking is necessary for this purpose.
    An example of application is to be able to automatically insert a vehicle 13 meters long and 3 meters wide, carrying a payload of the order of 60 tons, in a parking space with a positioning accuracy of l centimeter from a starting position at least fifteen meters or more from the parking space, parking place.
    Several more or less automated parking assistance solutions are known, but they have disadvantages or are unsatisfactory, or can not meet the problem.
    In the document EP 2 353 690 B1, it is the same context with the same type of vehicle, with the same types of areas of movement and parking. Nevertheless, the driving of the vehicle is completely manual and no specific travel requirement is mentioned.
    In US 8,862,321 B2 a vehicle is guided along an initial target path by controlling a steering actuator until the vehicle is in a particular desired position for parallel parking. Nevertheless, the vehicle considered is different from that treated by the present invention, it does not have the same dimensions or the same kinematics. The trajectory and the desired position are different. This approach does not meet the accuracy requirements.
    The document DE202013011728 raises the problem of the autonomous navigation of a vehicle in an external environment where the navigation of a container truck in a port environment is part of the possible application contexts. However, the solution described in this document is not suitable for complex motion kinematics. It offers markers set in the environment that indicate the path to follow with sensors on the vehicle that locate these markers. It has the particular disadvantage of requiring markers on the path of the vehicle, which complicates the implementation.
    An object of the invention is in particular to allow the parking of a vehicle in a docking station, fast and accurate. For this purpose, the subject of the invention is an automatic driving method for inserting a vehicle into a docking station and / or extracting said station to a target position, said vehicle being subjected to lateral sliding. and / or longitudinal of the front wheels and the rear wheels during its movements, said method comprising the following phases: a preliminary phase in which said vehicle is aligned to engage towards said target position; a first phase in which a trajectory to follow, called reference, is generated as a function of the state of the vehicle and of the target position, said state being defined by the current position and the orientation of the vehicle; a second phase in which, said reference trajectory being cut into sections, it is predicted at the start of each section and before the vehicle initiates a movement if said reference trajectory can be tracked as a function of imposed space constraints and slips lateral and / or longitudinal estimates; in a third phase, if said trajectory can be followed, the steering angle of said wheels and the linear traction speed of the vehicle are controlled according to the state of the vehicle and the lateral and / or longitudinal slippage, in order to rally the centers of said wheels on the reference path; if the trajectory can not be followed, the vehicle is re-aligned to the target position and a new reference trajectory is generated according to the first phase.
    In a particular implementation, several geographic areas are defined between the initial position of the vehicle and the target position in which said imposed constraints are increasingly severe when approaching said docking station.
    The control law of the steering angles βί, βν of the front and rear wheels is for example obtained according to a process of optimizing a function under stress where the variable is a vector u composed of derivatives with respect to the distance β'β , β'ν steering angles, regardless of time.
    Advantageously, the control law is for example based on a kinematic model of said vehicle taking into account the difference df, dr between the axis of the wheels and the steering axis at the front and rear of said vehicle.
    A vector u is for example calculated for each section of said trajectory.
    The constraints are for example a function of said imposed constraints depending on the size of said vehicle. The state of said vehicle and the slips are for example derived from an observation of variables independent of said trajectory, said variables being: the average rotational speeds and ωτ of the front and rear wheels; the steering angles of said front and rear wheels ay, ar; derivatives with respect to the time of said angles; the position (xm, ym) and the angle (0m) of a movable marker (m) linked to said vehicle relative to a fixed marker (o).
    The vehicle is for example a container carrier, the docking station being intended for loading and unloading containers, said container carrier having elevation movements causing a difference df, dr between the axis of the wheels and the axis of the container. direction to the front and back. The invention also relates to an automatic driving control device for inserting a vehicle into a docking station and / or extracting said station to a target position, said vehicle being subjected to lateral sliding and and / or longitudinal of the front wheels and the rear wheels during these movements, said device being adapted to be embedded in said vehicle and to be connected at least, via appropriate interfaces, to proprioceptive sensors, exteroceptive sensors and engines. actuation of the direction and traction of said vehicle, and comprising a computer implementing the method as described above. Other characteristics and advantages of the invention will become apparent with the aid of the description which follows, given with regard to appended drawings which represent: FIGS. 1a and 1b, an example of application of the invention; - Figures 2a and 2b, more specific views of the application showing respectively a vehicle parked in a docking station and engaging in the station; - Figure 3, the vehicle in the approach phase: - Figure 4, a partial view, a feature of a vehicle to which the invention applies; FIG. 5, a functional architecture of the control of a vehicle, according to the invention; FIG. 6, the phases of the method according to the invention; - Figure 7, an illustration of defined geographical areas which correspond to the constraints imposed on the vehicle; FIG. 8, an example of a kinematic model used by the invention.
    FIGS. 1a and 1b show an example of application of the invention for a container carrier. Figure 1a shows a mobile platform 1 towed or pushed by a motorized cabin 2. In the example of Figure 1a, the trailer 1 is for example 13 meters long and 3 meters wide. It is used in port traffic to be loaded with a container whose weight reaches several tens of tons.
    Figure 1b shows a parking station 3 for receiving the trailer 1. The parking station will be described in more detail later. In the example of Figure 1, it supports a container 4 ready to be loaded on the trailer 3 when it is parked in the station 3. The invention is presented with an example of a particular parking area 3. It applies of course for other types of parking areas. More generally, it also applies to precise parking assistance, especially long vehicles, apart from port traffic.
    Figures 2a and 2b specify the particular example of application of the invention mentioned in Figures 1a and 1b. Figure 2a shows in a view from above the trailer 1 parked in the parking station 3, which station can be called thereafter loading station, meaning that the trailer is loaded or unloaded from its container in this station.
    FIG. 2b presents, in a partial perspective view, the trailer 1 entering the station 3. This station comprises two side walls 21, 22 in the form of crenellations. The slots correspond to posts 23 surmounted by support 24, projecting towards the inside of the walls. These supports 24 are intended to carry together a container, the set of supports forming a reception surface. In the example of Figure 2a, three posts 23 are arranged on each side.
    The trailer 1 has a series of brackets 25 protruding outwardly arranged laterally on each side. In parking, as illustrated in Figure 2a, the supports 25 of the trailer are arranged between the posts. The trailer is then ready to be loaded.
    The loading of a container 4 on the trailer 1 is as follows: - The trailer is properly parked as shown in Figure 2a; - The container 4 is placed on the supports 24 of the station as shown in Figure 1b; - The supports 25 of the trailer being fixed on a structure which lifts, the supports are lifted with the structure to reach the underside of the container, then again raised to exceed the level of the supports 24 of the station, the container then being worn by the supports 25 alone of the trailer; - When the supports 25 of the trailer have reached a given level, the trailer 1 can be removed from the station, loaded by the container, the carriers 25 carrying the container then passing above the supports 24 of the station.
    For the unloading of the container, the steps are carried out in reverse order.
    The preceding steps show that the trailer 1 must be placed precisely in the loading station 3, in particular facing the side walls 21, 22. The expected accuracy can be of the order of a centimeter or less. In addition to this requirement of precision, there is a time constraint, the trailer having to be parked quickly in station 3.
    FIG. 3 illustrates the vehicle 1 in the approaching phase towards the docking station 3. Initially the vehicle is parked in a zone defined with respect to the docking station, this being covered by the detection beam 30 a sensor placed on the vehicle, for example rear if the vehicle parks in reverse. The vehicle moves along a reference path by controlling steering and traction actuators until it is in a target position relative to the docking station.
    FIG. 4 presents, by a partial view, a feature of the kinematics of a vehicle 1 container carrier used in the context previously described. More particularly, Figure 3 describes the elevation system which allows to lift the supports 25 to place or deposit the containers on the supports 24 of the station. The elevation is achieved by means of a jack 31 coupled to a mechanical part 32 fixed relative to the trailer 1. An end of the jack is mechanically secured to this piece according to a degree of rotation. The jack is also coupled to a mechanical part 33 movable in rotation about the axis of the wheels, identified by a point B. The other end of the jack is mechanically linked to one end of the movable piece with a degree of freedom. The other end of the movable part 33 is mechanically connected to the fixed part 32 also according to a degree of freedom. All these mechanical components 31, 32, 33 are movable in rotation in the plane perpendicular to the axis of the wheels, that is to say in the vertical plane. The elevation is controlled by the stroke 35 of the piston 311 of the jack, this movement being controlled by a control device, not shown. The stroke of the piston in extension causes a tilting of the moving part 33 relative to the axis B causing the increase of the angle μ 'between the fixed part 32 and the moving part 33, and therefore the elevation of the part fixed 32. As shown in Figure 4, these movements of rotation and elevation cause a shift between the axis B of the wheels and the steering axis 34 of the trailer, this offset is noted df for the front wheels and will be called interaxle afterwards. The lifting mechanism and the center distance are the same type for the rear wheels, the center distance is noted dr. The existence of these distances df, dr can increase the size when the vehicle is turned, going against the accuracy required for parking. The invention takes into account this particular kinematics of an elevating type container-carrying vehicle whose distance between the axis of the axle of the wheels and the steering axis varies according to the height of elevation of the vehicle. vehicle, especially the elevation of the supports. This consideration of the particular kinematics allows a more precise trajectory tracking. For the implementation of the method according to the invention a controller, embedded on the vehicle, automatically generates a trajectory to follow with respect to the target position in the docking station.
    Figure 5 illustrates the control architecture of the trajectory followed by the vehicle. It comprises: the controller 51 using a kinematic model of the vehicle 1; at least one observer 52 of the lateral and longitudinal drifts, corresponding to the slips, from a kinematic or dynamic 2D model, allowing the correction in real time or "on line" of the errors due to these drifts; an estimator 53 estimating the shifts from observations provided by the observer 52.
    This architecture, based on a real-time update, facilitates the processing of low-speed movements without penalizing higher-speed displacement processing by slippage correction. Vehicle condition and slips are observed from this model.
    A control device according to the invention incorporates the controller 51, the observer 52 and the estimator 53, these elements being functions performed by a computer thus implementing the different phases of the method according to the invention.
    The control device, fixed on the vehicle, is connected, via appropriate interfaces, to: - a power source ensuring its power supply; a perception system which gives position and orientation information with respect to the docking station, this system being in particular described in the patent application FR 1455049. motors for actuating the direction and the traction the vehicle that the controller controls in steering angle and linear traction speed; - vehicle proprioceptive sensors: front and rear axle angle encoders, front and rear axle speed measurement encoders and a distance sensor indicating the vehicle's vertical position, position being for example indicated by the distance measurement between a support and the ground.
    FIG. 6 illustrates the phases of the method according to the invention implemented by an architecture of the type of FIG. 5, these phases being repeated all along the parking trajectory.
    In a first phase 61, a trajectory to follow, called reference, is generated as a function of the vehicle's target parking position in the docking station 3.
    In a next step 62, it is predicted before the movement of the vehicle, if this trajectory can be followed according to the imposed constraints. These constraints may be the congestion constraints imposed by the docking station or congestion constraints related to the movements of the vehicle itself. If this reference path does not reach the target parking position, it is abandoned and the vehicle engages a maneuver to line up again to the docking station. This maneuver can be controlled remotely or manually by a driver. We return to the first phase 61 and a new reference trajectory is then predicted.
    If the reference trajectory leads to the target parking position, in a third phase 63, the vehicle is automatically driven to join and follow the reference trajectory according to the observation of the vehicle state and the slips, in real time and according to a constrained optimization process that will be detailed later.
    The verification phases 62 and trajectory tracking 63 are performed regularly along the path of the vehicle according to a distance sampling which will be specified later.
    Figure 7 shows geographical areas 71, 72, 73 arranged in front of the docking station. The vehicle must pass through these geographical areas before committing to the station. These zones are characterized by increasingly strong constraints, in the process of optimization under constraint of rallying to the reference trajectory, as one approaches the station.
    A first zone 71, furthest from the station, is intended for the alignment of the vehicle facing the station. The vehicle can be placed in this area non-automatically, for example manually by a driver. The initial trajectory of the vehicle is aligned with the entrance to the station. From this zone 71, the controller estimates the final position of the vehicle in the station according to the estimated trajectory.
    A second zone 72, called the approach zone, follows the preceding one with greater congestion constraints on the positioning. In other words, in this zone 72 we accept lower positioning errors with respect to the previous zone.
    A third zone 73, called the adjustment zone, is located just in front of the entrance to the station. Positioning errors must still be lower.
    Throughout the movement of the vehicle within these areas, the final position is estimated and the controller can therefore verify that the vehicle will be able to reach its target position relative to the station without risk of collision with the station. this. For this, congestion constraints are defined according to the area where it is located, and the controller anticipates their compliance along the path to follow to the target position. If it is impossible to reach the desired position or not detection of the docking station 3 by the vehicle, a signal is for example sent by the controller for manual modification of the vehicle position, by a driver or by remote control.
    FIG. 8 shows the 2D kinematic model of the vehicle mentioned above, opposite a calculated trajectory 83 constructed according to a predictive model and the current position of the vehicle. It is used by controller 51. This is an extension of a model described in the A. Micaelli &amp; C.Samson "Trajectory tracking for unicycle-type and two-steering-wheels mobile robots", Research Report RR-2097, INRIA, 1993, adapted to Figure 8.
    In particular the model of Figure 8 takes into account the distances df, dr front and rear between the steering axis and the wheel axis. Taking these distances into account in the vehicle model improves the accuracy of trajectory tracking.
    For reasons of simplification, a single front wheel 81 represents all of the two front wheels and their axis. Similarly, a single rear wheel 82 represents the set of two rear wheels and their axis. A local mark (/) is linked to the front wheels and a local mark (r) is linked to the rear wheels. The control of the trajectory of the vehicle is operated on these marks (/), (r) which must connect the calculated reference trajectory 83.
    For reasons of clarity, the center of the local coordinate system (/) has not been represented on the trajectory 83, at the current point P. It is the same for the reference (r).
    The front wheels 81 have a velocity vector Vf making an angle 8f with respect to the steering axis of the wheels. This angle 8f represents the sliding of the wheels on the ground. The velocity vector Vf makes an angle θf with the axis 80 of the vehicle, equal to the steering angle θ corrected by the slip angle θf. It is the same at the rear wheels 82 respectively for the velocity vectors Vr and angles 8r, / R, ar. The angles βf and / R express the steering angles of the front and rear wheels.
    The resulting velocity vector V of the vehicle makes an angle β with respect to the axis 80 of the vehicle. This angle β expresses the resulting slip of the vehicle, in the absence of slip β = 0.
    The direction of the axis 80 of the vehicle is marked in an absolute reference (o) by an angle 0m. This angle 0m also represents the angle between the movable marker (m) linked to the vehicle and the absolute reference (o).
    The center of the front wheel 81 makes a difference y ^, or lateral error, with the reference trajectory 83. More precisely, by the under-constrained optimization process, this difference must be reduced so that the center of the wheel 81 reaches the trajectory at a point P where a tangential local coordinate mark Cf makes an angle QCf with the movable marker (o), this angle 9Cf being the angle of the reference trajectory at point P. Similarly, the center of the rear wheel 82 must join an unrepresented point of the trajectory 83. From the model illustrated in FIG. 8, it is possible to deduce a linear model that will be used to generate the reference trajectory 83 for parking the vehicle. Linearization is performed around the steady state defined by: ~ @cf> - Vf ~ yr ~ Vr being the difference corresponding to the gap for the rear wheel; - β / = βτ = 0.
    The linearization is thus carried out around a state where the wheel is positioned on the trajectory 83 (the lateral errors y ^, yr being zero) and where the sliding and wheel angles are impaired, meaning that the direction and the angle of steering are damaged relative to the axis 80 of the vehicle.
    We consider the state vector
    y'f and y'r respectively being the derivative, with respect to the curvilinear distance, of the lateral error before yf and of the rear lateral error yr.
    We also consider the stall vector
    It can be shown that there are matrices A, P and B leading to the linearized model defined by the following relations: y = Px (1)
    (2) where y 'is the derivative, with respect to the curvilinear distance of y, β'f and β'ν being respectively the derivatives, with respect to the curvilinear distance, of βί and / R, the input vector
    thus expressing the evolution of the steering angles of the front and rear wheels according to the movement of the vehicle.
    The matrices A, P and B incorporate the equations of the kinematic model illustrated in FIG. 8 taking into account the distances df, dr.
    For the observation of the state of the vehicle, it is possible for example to use the vector:
    indicating the position and the orientation of the vehicle with respect to the absolute reference (o), where 6m is the angle of the mobile reference mark (m), related to the vehicle, with respect to the reference (o) as indicated above, where xm and ym are the coordinates of the movable marker (m) in the reference (o).
    For slip observations, the speeds and positions of the local marks (f) and (r) respectively related to the front and rear wheels are used. The speed of rotation of the wheels will be more particularly used to observe the longitudinal slips.
    On the trajectory 83, we define a number nsiep of sections, or not, curvilinear of length S. From the linearized model of relations (1) and (2), the state matrices of the continuous system are Ac = AP-1 and Bc = B. By discretizing a step of length S along the trajectory, the matrices of the system then become Ad and Bd given by the system of following equations (3):
    Where U is the identity matrix of dimension 4 x 4. A power calculation on Ac shows that Ac3 is equal to 0. For considerations of displacement in forward and reverse, the calculations are made for S> 0 and S <0.
    The predicted states of the vectors are calculated from the knowledge of the current state yo and the matrices Ad and Bd.
    The prediction on nsiep not successive leads to the following relation (4):
    (4) speaking more synthetically according to the following relation:
    (5)
    Relationships (4) and (5) show that at each step of distance k, an input command Uk is defined.
    In this constrained optimization process to follow the trajectory 83, the criterion to be optimized is for example a quadratic function of the predicted state vector Y and the control input U. This criterion, written crit, is expressed according to the following relation:
    (6) where Q and J1 are the 4nstep x 4nstep and 2 nstep x 2 nstep dimensional weight matrices respectively for vehicle condition and control. From relations (5) and (6), the criterion can be expressed according to the following relation (7):
    (7) where the parameter dependent on yo is expressed separately.
    Expressions of state constraints for a given vehicle are defined by relation (8) below. More particularly, according to a first approximation, the vehicle is modeled as a rectangle whose front side and the rear side are respectively at a distance Df and from the front axle along the longitudinal axis 80. The fact that these front side and back remain in a tolerance range Sgap around the trajectory can be expressed according to the following relation:
    (8)
    The tolerance range δgap is more and more restrictive depending on the areas 71, 72, 73, 3, the closer to the target parking position.
    This constraint can still be written in the following matrix form:
    (9)
    Extended to all nstep predicted states and written as a function of future control entries, this expression becomes:
    (10) where: • © is a matrix of nstep blocks arranged diagonally, each block consisting of a matrix D according to relation (9). "
    Obtaining the control input vector U then results from the solution to the following system where the first member is relative to the criterion to be optimized and the second member to the constraint:
    (11)
    Advantageously, the control law is defined independently of time and speed.
    To pass the real-time control law of the actuators, it is then necessary to involve the time. Extract first element of the U solution, gives the control law:
    (12)
    Vf being the linear speed of the front wheels.
    This gives a conventional system of optimization under stress.
    The control law defined by the relation (12) gives the derivative of the steering angles with respect to time, and thus indicates to the controller the steering commands to be applied. This information is then transmitted via interfaces adapted to impulse the movements to the wheels.
    Advantageously, the observation function is independent of the trajectory to follow and therefore not affected by tracking errors. She makes three types of observations. These observations allow correction of the lateral and longitudinal drifts with respect to the reference trajectory, in real time.
    A first type of observation makes the observation of the state of the vehicle defined in particular by its position, orientation and speed. This state can be obtained conventionally by extended Kalman filtering.
    A second type of observation observes the longitudinal slips of the vehicle and more particularly the front and rear wheels. In practice, this observation can be made by observing the apparent radius of the wheels, the apparent radius being calculated as a function of the distance traveled taking into account the longitudinal slips. The apparent radius is larger or smaller than the actual radius of the wheels, except in the absence of slip where it is equal. The comparison between the apparent radius and the real radius thus makes it possible to observe longitudinal slippage.
    A third type of observation observes lateral slides 8f, 8r illustrated in FIG.
    For these different observations, proprioceptive state variables, specific to the vehicle, are measured, in particular: the average rotation speeds oif and ωτ of the front and rear wheels; the forward and reverse steering angles ay, ar illustrated in FIG. 5; the derivatives with respect to the time ay, at these angles, that is to say the angular velocities of direction.
    The measured exteroceptive variables include: - the components of the vector
    indicating the position and orientation of the vehicle with respect to the absolute reference (o);
    The difference between the measurements of the proprioceptive variables, indicating the theoretical state of the vehicle, and the measurements of the exteroceptive variables, taking account of the slips of the vehicle, makes it possible to obtain the observation of the state of the vehicle, for example by a filter of extended Kalman. From these observations, the longitudinal and lateral sliding can be estimated, as illustrated in particular by the slip estimator 53 using the observation data 52 in the functional architecture illustrated in FIG. 5. The invention has been described. for a truck or container trailer type vehicle. It applies for all types of vehicles requiring automatic driving to a target position, subject to longitudinal and / or lateral sliding during their movements to this target position.
    权利要求:
    Claims (9)
    [1" id="c-fr-0001]
    1. A method of automatic driving for the insertion of a vehicle into a docking station (3) and / or extraction of said station to a target position, said vehicle being subjected to lateral and / or longitudinal sliding of front wheels (81) and rear wheels (82) during its movements, characterized in that it comprises the following phases: - a preliminary phase wherein said vehicle is aligned to engage said target position; a first phase (61) in which a trajectory (83) to be followed, referred to as a reference, is generated as a function of the state of the vehicle and of the target position, said state being defined by the current position and the orientation of the vehicle; a second phase (62) in which, said reference trajectory being cut into sections, it is predicted at the start of each section and before the vehicle engages a movement if said reference trajectory can be followed as a function of imposed congestion constraints and estimated lateral and / or longitudinal slips; in a third phase (63), if said trajectory can be followed, the steering angle of said wheels (81, 82) and the linear speed of traction of the vehicle are controlled according to the state of the vehicle and the lateral sliding and / or longitudinal, to join the centers of said wheels on the reference path (83); if the trajectory can not be followed, the vehicle is re-aligned to the target position and a new reference trajectory is generated according to the first phase (61).
    [2" id="c-fr-0002]
    2. Method according to claim 1, characterized in that defines several geographical areas (71, 72, 73, 3) between the initial position of the vehicle and the target position in which said imposed constraints are increasingly severe in s approaching said docking station.
    [3" id="c-fr-0003]
    3. Method according to any one of the preceding claims, characterized in that the control law of the steering angles βί, βν of the front and rear wheels is obtained according to a process of optimizing a function under stress where the variable is a vector u composed of the derivatives with respect to the distance β'β'Γ of the steering angles, independently of the time.
    [4" id="c-fr-0004]
    4. Method according to claim 3, characterized in that said control law is based on a kinematic model of said vehicle taking into account the difference (df, dr) between the wheel axis (B) and the steering axis (34) at the front and rear of said vehicle.
    [5" id="c-fr-0005]
    5. Method according to any one of claims 3 or 4, characterized in that a vector u is calculated for each section of said trajectory.
    [6" id="c-fr-0006]
    6. Method according to any one of claims 3 to 5, characterized in that the constraints are a function of said imposed constraints depending on the size of said vehicle.
    [7" id="c-fr-0007]
    7. Method according to any one of the preceding claims, characterized in that the state of said vehicle and the slips are derived from an observation (52) variables independent of said trajectory, said variables being: - the average rotational speeds a) f and or front and rear wheels; - the steering angles of said front and rear wheels ay ·, ar; derivatives with respect to time at said angles; the position (xm, ym) and the angle (0m) of a movable marker (m) linked to said vehicle relative to a fixed marker (o).
    [8" id="c-fr-0008]
    8. Method according to any one of the preceding claims, characterized in that the vehicle is a container carrier, the docking station being intended for loading and unloading containers, said container carrier having elevation movements causing a gap (df, dr) between the wheel axis (B) and the steering axis (34) at the front and rear.
    [9" id="c-fr-0009]
    9. Automatic driving control device for the insertion of a vehicle into a docking station (3) and / or extraction of said station to a target position, said vehicle being subjected to lateral sliding and / or longitudinal of the front wheels (81) and the rear wheels (82) during these movements, characterized in that said device being adapted to be embedded in said vehicle and to be connected at least, via appropriate interfaces, to proprioceptive sensors , exteroceptive sensors and motors for actuating the direction and traction of said vehicle, it comprises a computer (51, 52, 53) implementing the method according to any one of the preceding claims.
    类似技术:
    公开号 | 公开日 | 专利标题
    EP3152097B1|2019-04-03|Parking assistance device and vehicle provided with such a device
    EP3297876B1|2021-07-07|Movable conveyors for moving a four-wheel vehicle
    EP3115280B1|2018-04-18|Automatic driving method for inserting and removing a vehicle in a docking station, and monitoring device implementing such a method
    FR3072069B1|2019-09-20|METHOD FOR AUTOMATICALLY DRIVING A VEHICLE, IN PARTICULAR A BUS IN A STORAGE CENTER, AND DEVICE IMPLEMENTING SAID METHOD
    FR2948618A1|2011-02-04|Device for fitting different vehicles i.e. changing packs of batteries, in automobile field, has adjustable stop unit i.e. stop pin, adjusted to place component of vehicle in transversal and longitudinal positions to intervention zone
    CN106494400B|2020-10-23|Method for transferring goods from a transferring vehicle to a receiving vehicle and transferring vehicle
    EP3371035B1|2019-10-16|Device to assist with manoeuvres for parking alongside a platform
    FR3073767A1|2019-05-24|AUTOMATIC VEHICLE TRANSPORT SYSTEM WITH A TRANSPORT ROBOT
    EP2236445A1|2010-10-06|Device for stopping a truck, method and use of the same
    FR2969098A1|2012-06-22|METHOD AND DEVICE FOR ASSISTING THE DRIVER OF A VEHICLE
    US20200180695A1|2020-06-11|Parking robot for a transportation vehicle and method for operating such a parking robot
    EP3013721B1|2017-09-06|Method for steering an airport vehicle
    FR3048113B1|2019-06-14|DEVICE AND METHOD FOR AIDING THE DRIVING OF A MOTOR VEHICLE
    FR3046772B1|2019-06-14|MAINTENANCE DEVICE FOR THE MAINTENANCE OF RAILWAY VEHICLES
    EP3252441A1|2017-12-06|Weighing device and system for determining the mass and the centre of gravity of an aircraft
    EP1043262A1|2000-10-11|Method for loading a container on a platform of a transport means and installation for stocking and handling of containers
    US11148698B2|2021-10-19|Materials handling cart
    US11008201B2|2021-05-18|Automated rolling-jack for drive-on lifts
    FR3073880A1|2019-05-24|VEHICLE TRANSPORT SYSTEM AND METHOD FOR IMPLEMENTING THE SAME
    WO2020030857A1|2020-02-13|Electronic system for controlling the docking of a vehicle with a docking area, and corresponding method
    FR3096947A1|2020-12-11|A method of following the spiral trajectory of a vehicle under constraints
    FR3098836A1|2021-01-22|Four-wheel vehicle transporter
    FR3106560A1|2021-07-30|System for determining the angular position of a vehicle with two pivot points
    WO2006100386A1|2006-09-28|Method for guiding a robot and corresponding device
    同族专利:
    公开号 | 公开日
    EP3115280B1|2018-04-18|
    US20170003686A1|2017-01-05|
    US9933785B2|2018-04-03|
    FR3038279B1|2017-07-21|
    EP3115280A1|2017-01-11|
    引用文献:
    公开号 | 申请日 | 公开日 | 申请人 | 专利标题
    US20060111820A1|2003-05-19|2006-05-25|Daimlerchrysler Ag|Vehicle control system|
    DE10350923A1|2003-10-31|2005-05-25|Daimlerchrysler Ag|Positioning method for positioning the trailer of an articulated goods vehicle beneath a container supported on a mounting frame, whereby the path to be taken by the trailer is automatically determined up until a target position|
    EP2181898A1|2007-09-03|2010-05-05|Sanyo Electric Co., Ltd.|Driving assistance system and connected vehicles|FR3072069A1|2017-10-10|2019-04-12|Commissariat A L'energie Atomique Et Aux Energies Alternatives|METHOD FOR AUTOMATICALLY DRIVING A VEHICLE, IN PARTICULAR A BUS IN A STORAGE CENTER, AND DEVICE IMPLEMENTING SAID METHOD|FR4306M|1965-01-11|
    US7715953B2|2003-04-25|2010-05-11|Glimpse Waters, Inc.|Trailer backing up device and method|
    US7154385B2|2004-11-12|2006-12-26|General Motors Corporation|Vehicle-trailer backing up system using active front steer|
    DE102008059830A1|2008-12-01|2010-06-10|Gottwald Port Technology Gmbh|Load transfer process in a warehouse for containers, in particular standard containers|
    DE102010003980A1|2010-01-01|2011-07-07|INFILTEC GmbH, 67346|Filter for water treatment|
    US9937953B2|2011-04-19|2018-04-10|Ford Global Technologies, Llc|Trailer backup offset determination|
    US9238483B2|2011-04-19|2016-01-19|Ford Global Technologies, Llc|Trailer backup assist system with trajectory planner for multiple waypoints|
    US9708000B2|2011-04-19|2017-07-18|Ford Global Technologies, Llc|Trajectory planner for a trailer backup assist system|
    DE202013011728U1|2012-12-06|2014-04-01|Hans-Heinrich Götting|Road steering module|
    US8862321B2|2012-08-15|2014-10-14|GM Global Technology Operations LLC|Directing vehicle into feasible region for autonomous and semi-autonomous parking|
    US9446713B2|2012-09-26|2016-09-20|Magna Electronics Inc.|Trailer angle detection system|
    US20150115571A1|2013-10-24|2015-04-30|GM Global Technology Operations LLC|Smart tow|
    FR3021938B1|2014-06-04|2016-05-27|Commissariat Energie Atomique|PARKING ASSIST DEVICE AND VEHICLE EQUIPPED WITH SUCH A DEVICE.|
    US9522699B2|2015-02-05|2016-12-20|Ford Global Technologies, Llc|Trailer backup assist system with adaptive steering angle limits|CN107878453B|2017-11-07|2019-07-30|长春工业大学|A kind of automobile emergency collision avoidance integral type control method for hiding dynamic barrier|
    DE102018215982A1|2018-09-19|2020-03-19|Zf Friedrichshafen Ag|Device and method for controlling a vehicle for a swap body|
    法律状态:
    2016-07-29| PLFP| Fee payment|Year of fee payment: 2 |
    2017-01-06| PLSC| Search report ready|Effective date: 20170106 |
    2017-07-31| PLFP| Fee payment|Year of fee payment: 3 |
    2018-07-27| PLFP| Fee payment|Year of fee payment: 4 |
    2020-04-10| ST| Notification of lapse|Effective date: 20200306 |
    优先权:
    申请号 | 申请日 | 专利标题
    FR1501414A|FR3038279B1|2015-07-03|2015-07-03|AUTOMATIC DRIVING METHOD FOR INSERTING AND EXTRACTING A VEHICLE IN A RECEPTION STATION, AND CONTROL DEVICE USING SUCH A METHOD|FR1501414A| FR3038279B1|2015-07-03|2015-07-03|AUTOMATIC DRIVING METHOD FOR INSERTING AND EXTRACTING A VEHICLE IN A RECEPTION STATION, AND CONTROL DEVICE USING SUCH A METHOD|
    EP16174546.8A| EP3115280B1|2015-07-03|2016-06-15|Automatic driving method for inserting and removing a vehicle in a docking station, and monitoring device implementing such a method|
    US15/183,646| US9933785B2|2015-07-03|2016-06-15|Automatic control method for the insertion and the extraction of a vehicle into and from a receiving station, and control device implementing a method of this kind|
    [返回顶部]