Search

International audience; — This paper deals with the navigation of a mobile robot in unknown environment. The robot has to reach a final target while avoiding obstacles. It is proposed to break the task complexity by dividing it into a set of basic tasks: Attraction to a target and obstacle avoidance. Each basic task is accomplished through the corresponding elementary controller. The activation of one controller for another is done according to the priority task. To ensure the overall stability of the control system, especially at the switch moments, properties of hybrid systems are used. Hybrid systems allow switching between continuous states in presence of discrete events. In this paper, it is proposed to act on the gain of the proposed control law. The aim is to ensure the convergence of a common Lyapunov function to all the controllers. This ensures the stability of the overall control. Simulation results confirm the theoretical study.

International audience; This paper addresses the problem of controlling the dynamics of parallel kinematic manipulators from a global point of view, where modeling, sensing and control are considered simultaneously. The methodology is presented through the examples of the Gough-Stewart manipulator and the Quattro robot.

International audience; Image-based visual servoing has been found generally satisfactory and robust in the presence of camera and hand-eye calibration errors. However, in some cases, singularities and local minima may arise. We propose a modification of the feature vector which alleviates this problem without need of introducing further information, i.e., only pixel coordinates and depth estimations are used. Using the task-function approach, we demonstrate the relationship between the velocity screw of the camera and the current and desired p oses of the object in the camera frame. Experimental results on a real robotic platform illustrate the presented approaches.

National audience

International audience; This article presents a real-time control system for an agricultural mobile machine (vehicle) based on an on-board vision system using a single camera. This system has been designed in order to help human beings in repetitive and di cult tasks in the agricultural domain. The aim of the robotics application concerns the control of the vehicle with regard to the reap limit detected in image space. The perception aspect in relation with the application has been described in previous work, and here we deal with the control aspect. We integrate image features issued from the modelling of the scene in the control loop, in order to perform image-based servoing technique. The vehicle behaviour described here concerns bicycle and neural models, and three control laws are then synthesized. The rst and the second are modelling approaches and use an interaction between the scene and the image space. They are based on regulation of a task function. The third is a \black-box modelling" technique and is based on a neural network. Finally, experimental results obtained with these di erent control laws in different conditions are presented and discussed.

International audience; We are interested i n A GV applications since a long time, particularly in the eld of car vehicles and agricultural machines. This paper deals with the design of robust lateral controllers expressed directly in the sensor space. We use the camera sensor to extract an image of the position and orientation of the vehicle in regard with the white band, and consider the modelling of the projected line in image space a s the sensor signal. The kinematic modelling of the vehicle takes into account a normalization with respect to the vehicle velocity. We analyse the behaviour of lateral controllers designed w i t h a p ole assignment technique in presence of pipeline delays in the closed loop and perturbations. We particularly consider the perturbations on the inclination angle of the camera and on the camera height. Due to the problem of oscillations and instability of this approach, we investigate a H 1 robust control technique. To validate these approaches we have built an 1/10 scale road and we simulate the behaviour of the vehicle with a cartesian robot. We give some experimental results obtained with our experimental site.

International audience; This paper details a vehicle detection and localization algorithm from a single perspective view. A video camera, embedded i n a car, provides a grey scale image of a lead car. This lead vehicle is equipped with three visual marks to make easier low and high level processes. The algorithm is designed for real time implementation. It is conceived f r om a direct 3D localization method c oupled with a tracking method b ased on a rigidity criteria of the trajectories of N points. The processes are implemented n a specialized h a r dware a r chitecture to satisfy real time constraints.

International audience; Working on new equipments to increase agricultural productivity and safety, reducing human action hardness , Cemagref and Lasmea have developed an help guidance system for agricultural vehicles. The aim of this system is to help the operator to guide the machine in order to give more attention to the order task. To r each this goal, we use a visual perception of the environment because this sensor gives a solution well adapted to lots of situations. This project began with researches on a visual guidance system for a lawn mower in 1990 and has been adaptated to the development of the same system for a combine harvester in 1995. We model the lawn mower with a virtual axle, use trajectory generation to guide it, and localize the vehicle at each iteration to insure the servoing task. For the combine harvester, we decide to use the task function approach 23] and then to control the combine harvester directly in the sensor space.

International audience; To address traffic saturation in cities, new “Urban Transportation Systems”, based on electric vehicles in freeaccess, are in developing. One necessary functionality of such systems is their ability to move in a platoon fashion. A global decentralized platoon control strategy, supported by inter-vehicle communications, is addressed in this paper, relying on nonlinear control techniques. The main interest in a global approach is that servoing error accumulation can be avoided, whatever the platoon length. However, absolute vehicle localization is then required. In urban applications, cameras are realistic sensors, but localization is supplied in a 3D visual virtual world, slightly distorted w.r.t. the actual metric one. To enable accurate guidance, local corrections to the visual world are here computed from the data supplied by a laser rangefinder mounted on the second vehicle, and then shared with the whole platoon. Full-scale experiments demonstrate the performance of the proposed approach.

International audience; Hybrid architecture for autonomous navigation robots is presented in this paper. The principal future aim of this architecture is to achieve multi-robot convoy which must navigate in urban background, where autonomous vehicles have to track the desired trajectory while avoiding possible obstacles (walkers, other robots, etc.). Currently, we consider only the case of fixed obstacles. We propose to coordinate two continuous controllers (trajectory tracking and obstacle avoidance) by discrete events. However, hard switch from obstacle avoidance to trajectory following controller, may cause collision or undesired effects due to nonholonomic constraint or switching effects. Therefore, we introduce a third controller to overcome these drawbacks.

[Departement_IRSTEA]EA [TR1_IRSTEA]TESAD / MOST; International audience; Les faibles conditions d'adhérence rencontrées par les véhicules évoluant en milieu naturel engendrent d'importants glissements qu'il est nécessaire de considérer pour concevoir des lois de commande performantes. L'objet de cet article est ainsi de proposer un algorithme de commande adapté à un véhicule à quatre roues directrices pour réaliser un suivi précis de trajectoire dans de telles conditions. Nous montrons tout particulièrement l'intérêt des quatre roues directrices pour contrôler non seulement l'écart latéral, mais également l'écart angulaire du véhicule par rapport à la trajectoire à suivre. La motivation de ces travaux concerne le développement des systèmes de guidage pour les engins agricoles demandant un suivi latéral et angulaire précis. Les applications privilégiées sont l'évolution en ligne dans les coteaux et en crabe sur terrain plat. Basée sur un modèle cinématique étendu, une commande non-linéaire adaptative est élaborée reposant sur un principe de backstepping en considérant successivement les commandes de direction des roues avant et arrière du véhicule. Les résultats expérimentaux, obtenus avec un véhicule à quatre roues directrices et équipé d'un unique capteur GPS centimétrique, démontrent les capacités et la robustesse de la loi de commande proposée.

Object grasping is a typical human ability which is widely studied from both a biological and an engineering point of view. This paper presents an approach to grasp synthesis inspired by the human neurophysiology of actionoriented vision. Our grasp synthesis method is built upon an architecture which, taking into account the differences between robotic and biological systems, proposes an adaptation of brain models to the peculiarities of robotic setups. The architecture modularity allows for scalability and integration of complex robotic tasks. The grasp synthesis is designed as integrated with the extraction of a 3D object description, so that the object visual analysis is actively driven by the needs of the grasp synthesis: visual reconstruction is performed incrementally and selectively on the regions of the object that are considered more interesting for grasping

In this article, we present an integrated manipulation framework for a service robot, that allows to interact with articulated objects at home environments through the coupling of vision and force modalities. We consider a robot which is observing simultaneously his hand and the object to manipulate, by using an external camera (i.e. robot head). Task-oriented grasping algorithms [1] are used in order to plan a suitable grasp on the object according to the task to perform. A new vision/force coupling approach [2], based on external control, is used in order to, first, guide the robot hand towards the grasp position and, second, perform the task taking into account external forces. The coupling between these two complementary sensor modalities provides the robot with robustness against uncertainties in models and positioning. A position-based visual servoing control law has been designed in order to continuously align the robot hand with respect to the object that is being manipulated, independently of camera position. This allows to freely move the camera while the task is being executed and makes this approach amenable to be integrated in current humanoid robots without the need of hand-eye calibration. Experimental results on a real robot interacting with different kind of doors are pre- sented.

National audience; Depuis le debut des annees 90, nous travaillons sur un systeme de contrôle temps reel applique au guidage d'engins agricoles par vision monoculaire. Il s'agit d'alleger les tâches de pilotage confiees aux conducteurs d'engins en proposant un co-pilotage par un systeme entierement automatique. Ces tâches concernent particulierement les travaux de recolte (moisson), de fenaison, et de labour. Dans cet article, nous nous interessons a l'aspect contrôle du vehicule par vision monoculaire a partir de la detection de l'interface fauche/non-fauche dans l'image. Le vehicule considere est une moissonneuse-batteuse. L'approche fonction de tâche introduite par Samson et Espiau a la n des annees 80 est utilisee. Un premier contrôleur est elabore pour les recoltes sur terrains plats. Un second est propose pour tenir compte des effets de la pente.

International audience; Since the last years, we are interested in visual servo-ing and we h a ve d e v eloped many visual servoing applications using 2D visual features. The aim of this paper is to study how w e c a n i n troduce 3D visual features in a robot control loop. We consider a camera mounted on the end e ector of the manipulator robot to estimate the pose of the target object. The required positioning task is to reach a speci c pose between the sensor frame and a target object frame. Knowing the target object model, we can localize the object in the 3D visual sensor frame and obtain the pose between the camera and the target object at each iteration. To insure the visual servoing task, we h a ve d e v eloped two approaches. The rst approach is based on the task function approach i n troduced by Samson and Espiau at the end of 80's. We considered a 3D visual sensor which gives the 3D visual pose of an object in the sensor frame. In this case, the sensor signal (dimen-sion 6) is composed of the position and the orientation of the frame object in the sensor frame. The second approach consists to split the trajectory between the current and the desired pose in a set of elementaries screwing tasks. At e a c h iteration, the situation between the sensor and the object is computed, and then the trajectory is updated.

International audience

L'hypothèse de roulement pur sans glissement est fréquemment utilisée pour le suivi de trajectoire de robots mobiles. Malheureusement, celle-ci n'est en général plus valide pour les véhicules évoluant en milieux naturels, et en particulier dans le cadre des applications agricoles. En effet, les faibles propriétés d'adhérence des terrains d'évolution nécessitent la prise en compte de ce phénomène dans la commande dédiée à ce type d'engin. En outre les propriétés géométriques et dynamiques des engins évoluant sur de tels terrains (inertie, garde au sol) ainsi que les actionneurs employés génèrent des phénomènes de retard augmentant les erreurs de suivi enregistrées lors de phases transitoires. Dans cet article plusieurs développements théoriques (basés sur des principes d'observation, de commande adaptative et prédictive), visant à conserver le même niveau de précision que celui enregistré dans le cas des véhicules routiers, sont étudiés et testés en vraie grandeur sur des engins agricoles.

This book comes as a result of the research work developed in the framework of two large international projects: the European Science Foundation (ESF) supported program NATEMIS (Nonlinear Acoustic Techniques for Micro-Scale Damage Diagnostics) (of which Professor Delsanto was the European coordinator, 2000-2004) and a Los Alamos-based network headed by Dr. P.A. Johnson. The main topic of both programs and of this book is the description of the phenomenology, theory and applications of nonclassical Nonlinearity (NCNL). In fact NCNL techniques have been found in recent years to be extremely powerful (up to more than 1000 times with respect to the corresponding linear techniques) in a wide range of applications, including Elasticity, Material Characterization, Ultrasonics, Geophysics to Maintenance and Restoration of artifacts (paintings, stone buildings, etc.). The book is divided into three parts: Part I - defines and describes the concept of NCNL and its universality and reviews several fields to which it may apply; Part II - describes the phenomenology, theory, modelling and virtual experiments (simulations); Part III -discusses some of the most relevant experimental techniques and applications.

Omnidirectional Vision-Based Control from Homography

International audience; This paper presents the use of 3D visual features in the so-called "Visual Servo-ing Approach". After having brie y recalled how the task function approach is used in visual servoing, we present the notion of 3D logical vision sensors which permit us to extract visual information. In particular, we a r e i n terested in those composed of the estimation of both a 3D point and a 3D attitude. We g i v e t h e c o n trol law expression with regard to the two kinds of visual features performed at video rate with our robotic platform. We present some of the experimental results and show the good convergence of the control laws.

Automatic guidance of urban electric vehicles relying on a sensory memory is addressed in this presentation. Two exteroceptive sensors are successively investigated in order to build the sensory memory: an RTK GPS sensor and a monocular camera. Then, nonlinear control laws, achieving an exact decoupling between lateral and longitudinal guidance functionalities is presented, so that path following and platoon control can be addressed separately. Inter-vehicles communication, relying on WiFi technology, is used in order to enable a global platoon control: each vehicle is aware on-line of the whole platoon configuration and not only of the preceding vehicle relative localization. Full scale experiments, carried out with Cycab vehicles, are finally reported, and demonstrate high accuracy vehicles control with any of the two exteroceptive sensors considered here

International audience; This paper presents homography-based visual servoing scheme valid for the entire class of central catadioptriccameras (including conventional perspective cameras). First geometrical relationships between imaged points and lines intwoviews are exploited to estimate a generic homography matrix from which a partial Euclidean reconstruction is obtained. The informations extracted from the homography and imaged points and lines are then used to design hybrid control laws.

National audience; Dans cet article, nous présentons un schéma d'asservisse-ment visuel en utilisant une nouvelle représentation de la projection des droites dans l'image fournie par une caméra omnidirectionnelle. En effet, la projection d'une droite de l'espace dans l'image d'une caméra catadioptrique, est une conique représentée par cinq coefficients alors que celle-ci ne possède que deux degrés de liberté. Afin d'obte-nir une représentation minimale nous proposons d'utiliser les droites polaires relatives à la projection des droites 3D dans l'image omnidirectionnelle. Nous définissons ainsi une caméra perspective virtuelle observant des droites. Nous donnons la forme analytique de la matrice d'interaction, liant les déplacements de la caméra catadioptrique aux variations des observations visuelles, dans les espaces nor-malisé et pixelique. Nous exploitons ensuite ce résultat pour construire une loi de commande adéquate. Des résultats de simulation et expérimentaux confirment la validité de notre approche.

[Departement_IRSTEA]EA [TR1_IRSTEA]TESAD / MOSTChapter 6 ISBN: 1-59454-594-4; International audience; The development of automatic guidance systems for agricultural vehicles is receiving a large attention from both researchers and manufacturers. The motivations in such automated devices are, on one hand, to reduce considerably the arduous driving task, and on the other hand, to improve the efficiency and the quality of the agronomic work carried out. Such guidance devices require realtime vehicle localization on an unstructured area, such as agricultural fields. Nowadays, RTK GPS sensor appears as a very suitable sensor for these applications, since it can supply this information with a satisfactory centimeter accuracy at a high 10 Hz frequency, without requiring any preliminary equipment of the field. In this chapter, it is demonstrated that very accurate curved path following can actually be achieved by agricultural vehicles, even on slippery ground, relying on a single RTK GPS sensor. In a first step, in order to benefit from recent advances in Control Theory, sliding effects have been omitted, and therefore guidance laws have first been designed relying on a vehicle kinematic model. More precisely, taking advantage from structural properties of these models (they can be converted into a so-called chained form), curved path following has been achieved by designing a non-linear control law. Full-scale experiments reveal a very satisfactory guidance accuracy, except when the vehicle enters into sharp curves or when it moves on sloping fields. In these two latter situations, guidance accuracy is damaged since the vehicle undergoes sliding effects. In agricultural applications, vehicle dynamic models appear untractable from the control design point of view. Therefore, it is here proposed to describe sliding effects as a structured perturbation acting on vehicle kinematic model. Adaptive Control framework, and more precisely Internal Model techniques, can then be used jointly with the above mentioned non-linear control law to reject sliding effects, when still preserving all the advantages of the previously designed non-linear guidance law. Experiments demonstrate satisfactory guidance accuracy when the vehicle moves along a slope or when it executes sharp curves, excepted at their beginning or end. These transient guidance errors mainly ensue from delays introduced by the actuation device. Since the shape of the path to be followed is known, beginning/end of curves or slopes can be anticipated. Model Predictive Control framework is here used to provide such an anticipation. Satisfactory experimental results display the performances of the overall control scheme.

International audience

International audience

[Departement_IRSTEA]EA [TR1_IRSTEA]TESAD / MOST; International audience; Automatic devices dedicated to vehicle guidance in off-road conditions are necessarily confronted with sliding phenomenon, since it may considerable damage the accuracy of the following task. Control laws taking explicitly into account such a phenomenon have already been designed in previous work. They can actually improve the guidance accuracy. However, their efficiency is highly dependent on the sliding parameters estimation (since these parameters cannot be provided by a direct measurement). In this paper, an observer-like estimator is designed, providing sideslip angles from a single exteroceptive sensor, namely a Real Time Kinematic GPS (RTK-GPS). Improvements in guidance accuracy, with respect to previous estimation approaches, is demonstrated through full scale experiments, addressing agricultural applications.

International audience

International audience

International audience

We present experimental results of a visual servoing approach to control an agricultural mobile machine. To detect the limit between mowed and unmowed natural zones, we use an original region segmentation algorithm based on a Markovian modeling of a set of sites. With the 2D informations extracted by the image processing algorithm, a visual servoing unit allows the machine to follow the detected limit. All these algorithms are implanted on a parallel architecture., Nous présentons une approche commande référencée vision (CRV) pour asservir une machine agricole. Pour détecter la limite entre une zone fauchée et une zone non fauchée, nous utilisons un algorithme original de segmentation en régions basée sur un modèle Markovien d'un ensemble de sites. A partir des informations 2D issues de l'algorithme de traitement d'images, un module basé sur la CRV permet à la machine de suivre la limite détectée. Tous ces algorithmes sont implantés sur une architecture parallèle.

International audience

International audience; This paper focuses on the benefits of using computer vision to control a Gough-Stewart parallel robot. Namely, it is recalled that observing the legs of such a mechanism with a calibrated camera, thus following the redundant metrology paradigm, simplifies the control law. Then, we prove in this paper that such a control law depends on a reduced set of kinematic parameters (only those attached to the geometry of the robot base) and that these parameters can be obtained by solving a linear system. Moreover, we show that the camera can be calibrated without much experimental effort, simply using images of the robot itself. All this means that setting up the control system consists only in placing the camera in front of the robot.

National audience; Le projet Omnibot a pour principaux objectifs, la conception et l'optimisation d'une boucle perception-commande, en partant du composant (imageur) jusqu'à la navigation du robot. Cette approche originale permet ainsi la création d'une nouvelle génération de capteurs pour la vision omnidirectionnelle tout en gardant une interaction constante avec des applications réelles. L'objectif final de ce projet est de proposer un démonstrateur mettant en œuvre un robot mobile command par un asservissement visuel omndirectionnel. Ce projet fait actuellement l'objet des quatre parties qui sont développées ci après.

International audience; In this paper we present a new approach to control the 6 degrees of freedom of a manipulator using the projection of 3D lines extracted from the image plane of an omnidi-rectional camera. The main motivation of this work is to overcome the problem of visibility when using classical vision sensor. Indeed one of the principal deficiency of classical visual servoing strategy is that some parts of the visual features may get out of the field of view during servoing leading to its failure. Panoramic cameras overcome this problem because of their wider field of view. The present paper is concerned with the use of lines in paracataioptric visual servoing. Indeed when dealing with real environments (in-door or outdoor) or industrial workpiece, the extraction and the tracking of lines is natural. We derive the panoramic image jacobian for lines in the case of image based visual servoing. The corresponding control law is then designed and validated.

International audience; In this paper we present a new approach to control the 6 degrees of freedom of a manipulator using the projection of 3D lines extracted from the image plane of an omni-directional camera. The main motivation of this work is to overcome the problem of visibility when using classical vision sensor. Indeed one of the principal deficiency of classical visual servoing strategy is that some parts of the visual features may get out of the field of view during servoing leading to its failure. Panoramic cameras overcome this problem because of their wider field of view. The present paper is concerned with the use of lines in omnidirectional visual servoing. Indeed when dealing with real environments (indoor or outdoor) or industrial workpiece, the extraction and the tracking of lines is natural. We derive the panoramic image jacobian for lines in the cases of image based and position based visual servoing. The corresponding control laws are then designed and validated.; Dans cet article nous présentons une stratégie de com-mande de robots manipulateurs à six degrés de liberté en utilisant comme entrées d'une boucle d'asservissement visuel des primitives relatives à la projection de droites dans le plan image d'une caméra panoramique (image catadioptrique). La motivation principal de ce travail est de répondre aux problèmes de visibilité lorsque des in-formations issues d'une caméra classique sont intégrées à une boucle de commande. En effet, une des princi-pales sources d'échec des schémas d'asservissement est relative aux sorties de champs des primitives visuelles. L'utilisation de caméras omnidirectionnelles permet de résoudre cette difficulté puisque leurs champs de vision est bien supérieur à celui des caméras classiques. Cet article concerne en particulier l'utilisation de la projection de droites dans l'image d'une caméra omnidirec-tionnelle comme entrée de la boucle de commande. Afin de réaliser la commande d'un système robotique par as-servissement visuel, il est nécessaire d'estimer la ma-trice jacobienne (matrice d'interaction) liant les mou-vements de la caméra aux mouvements des primitives visuelles dans l'image catadioptrique. Dans cet article, nous donnons la forme analytique de la matrice d'in-teraction pour la projection de droites dans l'image ca-tadioptrique permettant de construire une loi de com-mande adéquate.