Retour au site du LAAS-CNRS

Laboratoire d’analyse et d’architecture des systèmes

Publications de l'équipe gepetto

Choisir la langue : FR | EN

437documents trouvés

18286
20/05/2019

Dynamics Consensus between Centroidal and Whole-Body Models for Locomotion of Legged Robots

R.BUDHIRAJA, J.CARPENTIER, N.MANSARD

GEPETTO

Manifestation avec acte : IEEE International Conference on Robotics and Automation ( ICRA ) 2019 du 20 mai au 24 mai 2019, Montreal (Canada), Mai 2019, 7p. , N° 18286

Lien : https://hal.laas.fr/hal-01875031

Diffusable

Plus d'informations

Abstract

It is nowadays well-established that locomotion can be written as a large and complex optimal control problem. Yet, current knowledge in numerical solver fails to directly solve it. A common approach is to cut the dimensionality by relying on reduced models (inverted pendulum, capture points, centroidal). However it is difficult both to account for whole-body constraints at the reduced level and also to define what is an acceptable trade-off at the whole-body level between tracking the reduced solution or searching for a new one. The main contribution of this paper is to introduce a rigorous mathematical framework based on the Alternating Direction Method of Multipliers, to enforce the consensus between the centroidal state dynamics at reduced and whole-body level. We propose an exact splitting of the whole-body optimal control problem between the centroidal dynamics (under-actuation) and the manipulator dynamics (full actuation), corresponding to a rearrangement of the equations already stated in previous works. We then describe with details how alternating descent is a good solution to implement an effective locomotion solver. We validate this approach in simulation with walking experiments on the HRP-2 robot.

147016
18288
15/01/2019

The Pinocchio C++ library – A fast and flexible implementation of rigid body dynamics algorithms and their analytical derivatives

J.CARPENTIER, G.SAUREL, G.BUONDONNO, J.MIRABEL, F.LAMIRAUX, O.STASSE, N.MANSARD

GEPETTO

Manifestation avec acte : IEEE/SICE International Symposium on System Integration ( SII ) 2019 du 14 janvier au 16 janvier 2019, Paris (France), Janvier 2019, 6p. , N° 18288

Lien : https://hal.laas.fr/hal-01866228

Diffusable

Plus d'informations

Abstract

We introduce Pinocchio, an open-source software framework that implements rigid body dynamics algorithms and their analytical derivatives. Pinocchio does not only include standard algorithms employed in robotics (e.g. forward and inverse dynamics) but provides additional features essential for the control, the planning and the simulation of robots. In this paper, we describe these features and detail the programming patterns and design which make Pinocchio efficient. We also offer a short tutorial for easy handling of the framework.

145395
18101
01/01/2019

An overview of humanoid robots technologies

O.STASSE, T.FLAYOLS

GEPETTO

Ouvrage (contribution) : Biomechanics of Anthropomorphic Systems, Springer, N°ISBN 978-3-319-93870-7, Vol.281, N°310, Janvier 2019 , N° 18101

Lien : https://hal.laas.fr/hal-01759061

Diffusable

Plus d'informations

Abstract

Humanoid robots are challenging mechatronics structures with several interesting features. Choosing a humanoid robot to develop applications or pursue research in a given direction might be difficult due to the strong interdependence of the technical aspects. This paper aims at giving a general description of this interdependence and highlight the lessons learned from the impressive works conducted in the past decade. The reader will find in the annex a table synthesizing the characteristics of the most relevant humanoid robots. Without focusing on a specific application we consider two main classes of humanoid robots: the ones dedicated to industrial application and the ones dedicated to human-robot interaction. The technical aspects are described in a way which illustrates the humanoid robots bridging the gap between these two classes. Finally this paper tries to make a synthesis on recent technological developments 1. 1 Mechanical structure 1.1 General design principal Humanoid robots are complex mechatronic systems. As such, it is necessary to consider the the mechanical structure, the computational system and the algorithms as a whole and for a given application. The robot's size, weight and strength are important factors when designing its structure. Let us consider two general classes of applications: physical performances while doing motion generation and validation of biological and/or cognitive models. The ATLAS robot from Boston Dynamics is an example of the first category, while the Kenshiro robot [45] from Tokyo University is an example of the second category

144064
17172
01/12/2018

Multi-contact locomotion of legged robots

J.CARPENTIER, N.MANSARD

GEPETTO

Revue Scientifique : IEEE Transactions on Robotics, Vol.34, N°6, pp.1441-1460, Décembre 2018 , N° 17172

Lien : https://hal.laas.fr/hal-01520248

Diffusable

Plus d'informations

Abstract

Locomotion of legged robots on arbitrary terrain using multiple contacts is yet an open problem. To tackle it, a common approach is to rely on reduced template models (e.g. the linear inverted pendulum). However, most of existing template models are based on some restrictive hypotheses that limit their range of applications. Moreover, reduced models are generally not able to cope with the constraints of the robot complete model, like the kinematic limits. In this paper, we propose a complete solution relying on a generic template model, based on the centroidal dynamics, able to quickly compute multi-contact locomotion trajectories for any legged robot on arbitrary terrains. The template model relies on exact dynamics and is thus not limited by arbitrary assumption. We also propose a generic procedure to handle feasibility constraints due to the robot whole body as occupation measures, and a systematic way to approximate them using off-line learning in simulation. An efficient solver is finally obtained by introducing an original second-order approximation of the centroidal wrench cone. The effectiveness and the versatility of the approach is demonstrated in several multi-contact scenarios with two humanoid robots both in reality and in simulation.

147033
18495
01/12/2018

Motion Planning for an Elastic Rod using Contacts

O.ROUSSEL, P.FERNBACH, M.TAIX

GEPETTO

Rapport LAAS N°18495, Décembre 2018

Lien : https://hal.archives-ouvertes.fr/hal-01954894

Diffusable

Plus d'informations

Abstract

This paper focuses on the motion planning problem of an extensible elastic rod in collision-free or contact space. The rod is assumed to be handled by grippers either at one or both extremities. Furthermore, during manipulation, the grasped end may change. We show that the use of both quasi-static and dynamic models can be coupled efficiently with sampling-based methods. Sampling directly in the submanifold of static equilibrium and contact-free configurations allows to take advantage of the dynamic model to improve the exploration of the state space. In this way, thanks to the contact information (point, forces, direction, number of contacts), the exploration of the RRT approach can be improved. We present a new RRT-SLIDE algorithm which guides the roadmap extension with a sliding contact mode based on some principles of human reasoning. We show that our approach is probabilistically complete. We also demonstrate the necessity of considering contacts on complex scenarios with several simulation experiments. Besides its performances , our algorithm does not require further tuning phase for a new scenario. Note to Practitioners-This work was done under the industrial project Flecto (ANR-Digital Models). It aims at solving the assembly/disassembly task for a rod while satisfying the elasticity parameters of its material. For industrial applications, the resolution time is a critical point. On the one hand, probabilistic motion planning methods require to efficiently build a roadmap of valid rod configurations. On the other hand, accurate rod modeling implies the use of a simulator based on the finite element method. Nevertheless, the very large size of the roadmap, that leads to a high number of calls to the simulator, is conflicting with the high computational cost of finite-element based simulation. To overcome this problem, one solution is to reduce the number of simulator calls. This can be achieved by sampling the free space with an efficient parameterization and by limiting the use of the simulator to roadmap extension in the free space or in the contact space. We introduce heuristics based on contact information returned by the simulator to reduce significantly the computational time. One of the main advantages of our algorithm is that it does not require any tuning phase for each scenario. Although we do not solve the more general gripper manipulation planning problem, this approach could be used as a first step before computing the motion of the grippers. In the framework of our project, we did not consider disassembling operations implying undoing rod knots. Consequently, we do not take friction into account in our approach. In order to handle knots, it would be necessary to have a physics simulator that could handle friction for deformable rods.

146014
17381
27/11/2018

2PAC: Two point attractors for center of mass trajectories in multi contact scenarios

S.TONNEAU, A.DEL PRETE, J.PETTRE, N.MANSARD

GEPETTO, IRISA

Revue Scientifique : ACM Transactions on Graphics, Vol.37, N°5, 176p., Novembre 2018 , N° 17381

Lien : https://hal.archives-ouvertes.fr/hal-01609055

Diffusable

Plus d'informations

Abstract

Synthesizing motions for legged characters in arbitrary environments is a long-standing problem that has recently received a lot of attention from the computer graphics community. We tackle this problem with a procedural approach that is generic, fully automatic and independent from motion capture data. The main contribution of this paper is a point-mass-model-based method to synthesize Center Of Mass trajectories. These trajectories are then used to generate the whole-body motion of the character. The use of a point mass model often results in physically inconsistent motions and joint limit violations. We mitigate these issues through the use of a novel formulation of the kinematic constraints which allows us to generate a quasi-static Center Of Mass trajectory, in a way that is both user-friendly and computationally efficient. We also show that the quasi-static constraint can be relaxed to generate motions usable for applications of computer graphics (on average 83% of a given trajectory remain physically consistent). Our method was integrated in our open-source contact planner and tested with different scenarios-some never addressed before-featuring legged characters performing non-gaited motions in cluttered environments. The computational efficiency of our trajectory generation algorithm (under ten ms to compute one second of motion) enables us to synthesize motions in a few seconds, one order of magnitude faster than state-of-the-art methods.

145315
18650
21/11/2018

Modélisation et contrôle d'actionnaurs pour la robotique humanoïde

F.FORGET

GEPETTO

Doctorat : Université de Toulouse III - Paul Sabatier, 21 Novembre 2018, 147p., Président: P.DANES, Rapporteurs: C.CHEVALLEREAU, S.ALFAYAD, Examinateurs: S.CARON, Directeurs de thèse: N.MANSARD, O.STASSE , N° 18650

Lien : https://hal.laas.fr/tel-02128260

Diffusable

Plus d'informations

Abstract

Humanoid robotics actuation and control is complex as this field has specific needs. This involves the design of advanced mechatronic constructions. In this thesis, we focused on the humanoid robots actuation and paid particular attention to systems with good compliance capabilities. Compliance refers to the ability of the system to adapt to its environment by adopting a flexible behaviour. This feature is particularly sought after for human-robot interactions and to make contacts between the robot and its environment smoother. We studied two systems in detail : the first is a system based on Mckibben’s artificial pneumatic muscles and the second is an electric cable driven actuator with flexibility. We have developed and implemented a control architecture for the control of both systems. The interest of this architecture lies in its generic nature and in its use within a model predictive control scheme. We then studied the thermal behaviour of an actuator in order to exploit the full potential of this system while ensuring its integrity. Finally, we have carried out developments to improve the joint torque control of the humanoid robot Talos.

Résumé

La robotique humanoïde pose des problèmes d’actionnement et de contrôle complexes tant ce domaine a des besoins spécifiques. Cela implique la conception de constructions mécatroniques avancées. Dans le cadre de cette thèse, nous nous sommes intéressé aux actionneurs de ces robots et avons porté une attention particulière aux systèmes présentant de bonnes capacités de compliance. Par compliance, on entend la faculté du système à s’adapter à son environnement en adoptant un comportement souple. Cette caractéristique est particulièrement recherchée pour les interactions homme-robot et pour rendre les contacts entre le robot et son environnement plus doux. Nous avons étudié deux systèmes en détails : le premier est un système à base de muscles artificiels pneumatiques de Mckibben et le second, un actionneur électrique à câble comportant une flexibilité. Nous avons développé et mis en oeuvre une architecture de contrôle pour la commande des deux systèmes. L’intérêt de cette architecture réside dans son caractère générique et dans son exploitation au sein d’un schéma de commande prédictive (Model predictive control). Nous avons ensuite étudié le comportement thermique d’un actionneur en vue d’exploiter le plein potentiel de ce système tout en garantissant son intégrité. Finalement, nous avons mené des développements afin d’améliorer le contrôle en couple articulaire du robot humanoïde Talos.

Mots-Clés / Keywords
Robotique humanoide; Actionnement; Mécatronique; humanoid robotics; Actuation; Mecatronics;

147413
18345
09/11/2018

Cross-cultural (France and Japan) and Multidisciplinary Discussion on Artificial Intelligence and Robotics: Tendencies and Research Prospects

N.ABE

GEPETTO

Rapport LAAS N°18345, Novembre 2018, 24p.

Lien : https://halshs.archives-ouvertes.fr/halshs-01907840

Diffusable

Plus d'informations

Abstract

Artificial Intelligence (AI) is rapidly developing through a “Deep Learning model” that is based on a neural network and more powerful learning system than a traditional algorithm system. Our society is facing rapid technological changes that could lead to dramatic social impact in the economic, societal, educational and ethical landscape. In this context, the importance of multidisciplinary discussion and research in the field of Intelligent Systems (AI and Robot) is increasingly recognized. The present discussion paper aims to summarize the current state and research on Intelligent Systems and social impact and proposes multidisciplinary projects covering both engineering and social science studies. It presents three promising research topics that can provide us with a better understanding of Intelligent Systems and our attitude or relationship towards the technology: 1) international and national (France and Japan) initiatives with regard to the social impact of AI, 2) trust and acceptability in Intelligent Systems, and 3) Human-Machine Interaction.

Résumé

L'intelligence artificielle (IA) se développe rapidement grâce au « Deep learning », un modèle d'apprentissage automatique basé sur un réseau neuronal qui est plus puissant qu'un système algorithmique traditionnel. Notre société est confrontée à des changements technologiques rapides qui pourraient avoir un impact social dramatique dans les paysages économique, sociétal, éducatif et éthique. Dans ce contexte, l'importance de la discussion et de la recherche multidisciplinaires dans le domaine des systèmes intelligents (IA et robot) est de plus en plus reconnue. Le présent document vise à récapituler l'état actuel de la recherche sur les systèmes intelligents et leur impact social et à proposer des projets multidisciplinaires couvrant à la fois les études en ingénierie et en sciences sociales. Il présente trois thèmes de recherche prometteurs qui peuvent nous permettre de mieux comprendre les systèmes intelligents et notre attitude ou notre relation vis-à-vis de la technologie : 1) les initiatives internationales et nationales (France et Japon) concernant l'impact social de l'IA, 2) la confiance et l'acceptabilité des systèmes intelligents et 3) l'interaction homme-machine.

145045
18247
09/11/2018

Differential Dynamic Programming for Multi-Phase Rigid Contact Dynamics

R.BUDHIRAJA, J.CARPENTIER, C.MASTALLI, N.MANSARD

GEPETTO

Manifestation avec acte : IEEE-RAS International Conference on Humanoid Robots ( HUMANOIDS ) 2018 du 06 novembre au 09 novembre 2018, Beijing (Chine), Novembre 2018, 6p. , N° 18247

Lien : https://hal.archives-ouvertes.fr/hal-01851596

Diffusable

Plus d'informations

Abstract

A common strategy today to generate efficient locomotion movements is to split the problem into two consecutive steps: the first one generates the contact sequence together with the centroidal trajectory, while the second one computes the whole-body trajectory that follows the centroidal pattern. Yet the second step is generally handled by a simple program such as an inverse kinematics solver. In contrast, we propose to compute the whole-body trajectory by using a local optimal control solver, namely Differential Dynamic Programming (DDP). Our method produces more efficient motions, with lower forces and smaller impacts, by exploiting the Angular Momentum (AM). With this aim, we propose an original DDP formulation exploiting the Karush-Kuhn-Tucker constraint of the rigid contact model. We experimentally show the importance of this approach by executing large steps walking on the real HRP-2 robot, and by solving the problem of attitude control under the absence of external forces.

146704
18574
01/11/2018

Strategies of Parkour practitioners for executing soft precision landings

G.MALDONADO, P.SOUERES, B.WATIER

GEPETTO

Revue Scientifique : Journal of Sports Sciences, Vol.36, N°22, pp.2551-2557, Novembre 2018 , N° 18574

Lien : https://hal.laas.fr/hal-02042616

Diffusable

Plus d'informations

Abstract

Parkour landing techniques differ from performances of other sports as they are practiced in urban spaces with uncontrolled surfaces and drop heights. Due to the relatively young age of the sport, few studies have tried to understand how practitioners – called traceurs – succeed at performing these dynamic performances. In this paper, we focus on the precision landing technique, which has a fundamental role in most of the Parkour motions. We analyzed the lower limbs motion of traceurs executing the precision landings from two different heights and compared their performance with untrained participants. We found that traceurs perform a soft landing extending its duration twice than untrained participants do , increasing the range of motion and generating more mechanical energy to dissipate the impact. In the Parkour technique, the knee accounted for half of the energy dissipated. The peak joint torques and power were reduced in the Parkour technique. The increase of the landing height did not modify the proportion of individual joint mechanical energy contribution for dissipation. Our results could be used to enhance Parkour performance, and to understand new ways in which sport practitioners can land in order to prevent injuries.

146693
Les informations recueillies font l’objet d’un traitement informatique destiné à des statistiques d'utilisation du formulaire de recherche dans la base de données des publications scientifiques. Les destinataires des données sont : le service de documentation du LAAS.Conformément à la loi « informatique et libertés » du 6 janvier 1978 modifiée en 2004, vous bénéficiez d’un droit d’accès et de rectification aux informations qui vous concernent, que vous pouvez exercer en vous adressant à
Pour recevoir une copie des documents, contacter doc@laas.fr en mentionnant le n° de rapport LAAS et votre adresse postale. Signalez tout problème de dysfonctionnement à sysadmin@laas.fr. http://www.laas.fr/pulman/pulman-isens/web/app.php/