Proceedings of the AAAI Conference on Artificial Intelligence
In a recent position paper in Artificial Intelligence, we argued that the automated planning rese... more In a recent position paper in Artificial Intelligence, we argued that the automated planning research literature has underestimated the importance and difficulty of deliberative acting, which is more than just interleaving planning and execution. We called for more research on the AI problems that emerge when attempting to integrate acting with planning. To provide a basis for such research, it will be important to have a formalization of acting that can be useful in practice. This is needed in the same way that a formal account of planning was necessary for research on planning. We describe some first steps toward developing such a formalization, and invite readers to carry out research along this line.
Robotics is an interdisciplinary research field leveraging on control theory, mechanical engineer... more Robotics is an interdisciplinary research field leveraging on control theory, mechanical engineering, electronic engineering and computer science. It aims at designing machines able to perceive, move around and interact with their environment in order to perform useful tasks. Artificial Intelligence (AI) is an area of computer science, overlapping with but significantly distinct from robotics. Its purpose is to understand intelligence, through effective computational models, design and experiment with systems which implement these models. There is a significant convergence between Robotics and AI. Their intersection, qualified here as Machine Intelligence, is critical for both areas. Robots implement the so-called "perceptiondecision-action" loop; the intelligence or decision making part is central in that loop for tackling more variable and complex environments and tasks. On the other hand, AI is moving from abstract intelligence, such as in playing chess, to addressing embodied intelligence. This paper introduces the reader to some of the research issues and approaches in Machine Intelligence. It surveys the state of the art in key issues such as planning and acting deliberately on the basis of tasks and world models, learning these models, and organizing the sensory-motor and cognitive functions of a robot into resilient and scalable architectures.
Artificial systems are becoming more and more complex, almost as complex in some cases as natural... more Artificial systems are becoming more and more complex, almost as complex in some cases as natural systems. Up to now, the typical engineering question was "how do I design my system to behave according to some specifications". However, the incremental design process is leading to so complex artifacts that engineers are more and more addressing a quite different issue of "how do I model the observed behavior of my system". Engineers are faced with the same problem as scientists studying natural phenomena. It may sound strange for an engineer to engage in observing and modeling what a system is doing, since this should be inferable from the models used in the system's design stage. However, a modular design of a complex artifact develops only local models that are combined on the basis of some composition principle of these models; it seldom provides global behavior models.
International audiencePropriétés fondamentales du vivant, anticipation et prédiction sont au cœur... more International audiencePropriétés fondamentales du vivant, anticipation et prédiction sont au cœur de nos vies et de chacune de nos actions : on calcule les chances d’un homme politique d’être élu, les risques que l’avion dans lequel on monte s’écrase, on évalue les chances de bénéficier d’une météo favorable pendant nos vacances…Mais quels sont les contraintes, les mécanismes psychologiques et neurobiologiques qui jouent sur la capacité humaine d’anticipation ? Comment l’humain se projette-t-il dans le futur ? Quelles sont les logiques qui président à la mise en œuvre de cette capacité ? Quelles contributions la psychopathologie apporte-t-elle à la compréhension de ces mécanismes ? Comment l’homme réagit-il face à des situations qui l’obligent à reconfigurer son existence ?Telles sont les questions abordées dans cet ouvrage rédigé par un groupe de chercheurs – philosophes, physiologistes, neuropsychologues, mathématiciens, conseillers en entreprise – qui présente les multiples facettes de cette fascinante faculté humaine et ses pathologies, ainsi que son usage dans les processus de décision, que ce soit dans la vie quotidienne ou dans l’entreprise.Alain Berthoz, neurophysiologiste, est professeur émérite au Collège de France et membre de l’Académie des sciences. Il est l’auteur de nombreux ouvrages, dont Le Sens du mouvement, La Décision et La Simplexité, qui ont été de très grands succès.Claude Debru est philosophe, correspondant à l’Académie des sciences et professeur de philosophie des sciences à l’École normale supérieure de Paris. Il est l’auteur, avec Pierre Buser, de Le Temps, instant et durée.Contributions de Margherita Arcangeli, Daniel Bennequin, Pierre Bessière, Pierre Buser, Jérôme Dokic, Jacques Droulez, Malik Ghallab, Étienne Koechlin, Giuseppe Longo, Nicolas Morgado, Didier Naud, Richard Palluel-Germain, Denis Perrin, Jean-Luc Petit, Pascale Piolino, Armin Schnider
Au cours des dernieres annees, l'INRIA a connu un developpement sans precedent, confirmant so... more Au cours des dernieres annees, l'INRIA a connu un developpement sans precedent, confirmant son excellence au plan francais et europeen comme au plan international. En quatre decennies, l'Institut a non seulement su imposer une nouvelle discipline en France, l'informatique, mais aussi allier recherche fondamentale et reponses aux besoins de la societe.
150 research groups in Modeling, control and applied mathematics Programming, secure and re... more 150 research groups in Modeling, control and applied mathematics Programming, secure and reliable IT Communicating, networks, systems, grids Sensing and interacting, vision, graphics, VR and robotics A total work force of 3600 persons
A task planner for the computer-aided design of a space-lander robot
Proceedings. 1998 IEEE International Conference on Robotics and Automation (Cat. No.98CH36146)
The Rosetta project has to design, and launch a robotic comet lander in 2003 in order to carry ou... more The Rosetta project has to design, and launch a robotic comet lander in 2003 in order to carry out extensive scientific analyse's of the P/Wirtanen comet following a rendezvous in 2011. A mission simulator was developed in order to obtain information on the critical resources of the lander. However, one needs to feed a simulator with realistic task sequences, that
HAL is a multidisciplinary open access archive for the deposit and dissemination of scientific re... more HAL is a multidisciplinary open access archive for the deposit and dissemination of scientific research documents, whether they are published or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers. L'archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d'enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.
This paper introduces an original framework for planning and acting with hierarchical input/outpu... more This paper introduces an original framework for planning and acting with hierarchical input/output automata for systems defined by the parallel composition of the models of their components. Typical applications are, for example, in harbor or warehouse automation. The framework extends the usual parallel composition operation of I/O au-tomata with a hierarchical composition operation that can refine a task. It defines planning as the synthesis of a control component to drive, through I/O interactions and task refinement, the system toward desired states. A new nondeterministic algorithm performs this synthesis. We tackle these issues on a theoretical basis. We formally define the representation and prove that the two operations of parallel and hierarchical composition are distributive, which is essential for the correctness and completeness of the proposed planning algorithm.
Despite a very strong synergy between Robotics and AI at their early beginning, the two fields pr... more Despite a very strong synergy between Robotics and AI at their early beginning, the two fields progressed widely apart in the following decades. However, we are witnessing a revival of interest in the fertile domain of embodied machine intelligence, which is due in particular to the dissemination of more mature techniques from both areas and more accessible robot platforms with advanced sensory motor capabilities, and to a better understanding of the scientific challenges of the AI-Robotics intersection. The ambition of this paper is to contribute to this revival. It proposes an overview of problems and approaches to autonomous deliberate action in robotics. The paper advocates for a broad understanding of deliberation functions. It presents a synthetic perspective on planning, acting, perceiving, monitoring, goal reasoning and their integrative architectures, which is illustrated through several contributions that addressed deliberation from the AI-Robotics point of view. Robotics has always been a fertile inspiration paradigm for AI research, frequently referred to in its literature, in particular in the above topics. The early days of AI are rich in pioneering projects fostering a strong AI research agenda on robotics platforms. Typical examples are Shakey at SRI [85] and the Stanford Cart in the sixties, or Hilare at LAAS [36] and the CMU Rover [70] in the seventies. However, in the following decades the two fields developed in diverging directions; robotics expanded mostly outside of AI laboratories. Hopefully, a revival of the synergy between the two fields is currently being witnessed. This revival is due in particular to more mature techniques in robotics and AI, to the development of inexpensive robot platforms with more advanced sensing and control capabilities, to a number of popular competitions, and to a better understanding of the scientific challenges of machine intelligence, to which we would like to contribute here. This revival is particularly strong in Europe where a large number of groups is actively contributing to the AI-Robotics interactions. For example, out of the 260 members of the Euron network, 1 about a third investigate robotics decision and cognitive functions. A similar ratio holds for the robotics projects in FP6 and FP7 (around a hundred). Many other european groups not within Euron and projects outside of EU programs are equally relevant to the AI and Robotics synergy. This focused perspective on deliberative capabilities in robotics cannot pay a fair tribute to all european actors of this synergy. It illustrates however several contributions from a few groups throughout Europe. 2 Its ambition is not to cover a comprehensive survey of deliberation issues, and even less of the AI-Robotics intersection. In the limited scope of this special issue, we propose a synthetic view of deliberation functions. We discuss the main problems involved in their development and exemplify a few approaches that addressed these problems. This "tour d'horizon" allows us to advocate for a broad and integrative view of deliberation, where problems are beyond search in planning, and beyond the open-loop triggering of commands in acting. We hope through this perspective to strengthen the AI-Robotics synergies. The outline of the paper is the following: five deliberation functions are introduced in the next section; these are successively addressed through illustrative contributions; section 8 is devoted to architecture problems, followed by a conclusion. 2 Deliberation functions in robotics Deliberation refers to purposeful, chosen or planned actions, carried out in order to achieve some objectives. Many robotics applications do not require deliberation capabilities, e.g., fixed robots in manufacturing and other well-modeled environments; vacuum cleaning and other devices limited to a single task; surgical and other tele-operated robots. Deliberation is a critical functionality for
This position paper discusses the requirements and challenges for responsible AI with respect to ... more This position paper discusses the requirements and challenges for responsible AI with respect to two interdependent objectives: (i) how to foster research and development efforts toward socially beneficial applications, and (ii) how to take into account and mitigate the human and social risks of AI systems.
Structural elements inserted in proteins are essential to define folding/unfolding mechanisms and... more Structural elements inserted in proteins are essential to define folding/unfolding mechanisms and partner recognition events governing signaling processes in living organisms. Here, we present an original approach to model the folding mechanism of these structural elements. Our approach is based on the exploitation of local, sequence-dependent structural information encoded in a database of three-residue fragments extracted from a large set of high-resolution experimentally determined protein structures. The computation of conformational transitions leading to the formation of the structural elements is formulated as a discrete path search problem using this database. To solve this problem, we propose a heuristically-guided depth-first search algorithm. The domain-dependent heuristic function aims at minimizing the length of the path in terms of angular distances, while maximizing the local density of the intermediate states, which is related to their probability of existence. We ha...
Uploads
Papers by Malik Ghallab