Manipulating everyday objects without detailed prior models is still beyond the capabilities of existing robots. This is due to many challenges posed by diverse types of objects: Manipulation requires understanding and accurate model of physical properties of objects such as shape, mass, friction, elasticity, etc. Many objects are deformable, articulated, or even organic with undefined shape (e.g., plants) such that a fixed model is insufficient. On top of this, objects may be difficult to perceive, typically because of cluttered scenarios, or complex lighting and reflectance properties such as specularity or partial transparency. Creating such rich representations of objects is beyond current datasets and benchmarking practices used for grasping and manipulation. In this project we will develop an automated interactive perception pipeline for building such rich digitization. More specifically, in IPALM, we will develop methods for the automatic digitization of objects and their physical properties by exploratory manipulations. These methods will be used to build a large collection of object models required for realistic grasping and manipulation experiments in robotics. Household objects such as tools, kitchenware, clothes, and food items are not only widely accessible and in focus of many practical applications but also pose great challenges for robot object perception and manipulation in realistic scenarios. We propose to advance the state of the art by including household objects that can be deformable, articulated, interactive, specular or transparent, as well as shapeless such as cloth and food items. Our methods will learn physical properties essential for perception and grasping simultaneously from different modalities: vision, touch, audio as well as text documents such as online manuals and will include the following properties: 3D model, texture, elasticity, friction, weight, size and grasping techniques for intended use. At the core of our approach is a two-level modeling, where a category level model provides priors for capturing instance level attributes of specific objects. We will exploit online available resources to build prior category level models and a perception-action-learning loop will use the robot’s vision, audio, and touch to model instance level object properties. In return, knowledge acquired from a new instance will be used to improve the category-level knowledge. Our approach will allow us to efficiently create a large database of models for objects of diverse types, which will be suitable for example for training neural network based methods or enhancing existing simulators. We will propose a benchmark and evaluation metrics for object grasping, to enable comparisons of results generated with various robotics platforms on our database. The main objectives we pursue are commercially relevant robotics technologies, as endorsed by the support letters of several companies. We will pursue our goals with a consortium that brings together 5 world-class academic institutions from 5 EU countries (Imperial College London (UK), University of Bordeaux (France), Institut de Robòtica i informàtica Industrial (Spain), Aalto University (Finland), and the Czech Technical University (Czech Republic), assembling a complementary research team with strong expertise in the acquisition, processing and learning of multimodal information with applications in robotics.

The project is to develop the theory of homomorphisms of signed graphs whose study has begun in 2010 as a postdoc project of Reza Naserasr under the supervision of Éric Sopena. The theory is motivated by several extensions of the 4-Color Theorem: the Hadwiger conjecture, further extension of it (known as the odd-Hadwiger conjecture), the conjectures of Naserasr and Guenin (on mapping planar graphs to projective cubes), and two conjectures of Seymour (on determining edge-chromatic number of planar graphs and on characterizing binary clutters). A signed graph is a graph together with an assignment of signs to the edges. This assignment allows to apply a finer notion of minor where we are only allowed to contract positive edges, but we are also allowed to multiply signs of all edges incident to a same vertex by a negative sign. Thus, while the class of graphs with no triangle as a minor is the class of all forests, the class of fully negative signed graphs which do not contain fully negative triangle as a minor is the fully negative signed graphs built on any bipartite graph. Hence, while both forbidden structures imply 2-colorability of the input graphs, the latter applies to a larger family (in that example, exceptionally, it is a characterization). One of the main goals of our project is to develop that advantage in a bigger framework using the notion of homomorphisms of signed graphs. Of particular interest is the question of homomorphisms to signed projective cubes. These graphs are built from classic hypercubes by adding a negative edge between each pair of antipodal vertices. The existence of a homomorphism from a given signed graph to a signed projective cube is equivalent to a packing problem of the edges of the input graphs. A main question is the study of the mappings from (signed) planar graphs to (signed) projective cubes. The question is a direct extension of the 4-Color Theorem, it relates to the study of the circular and fractional chromatic numbers of planar graphs of given odd girth. It is also strongly related to determining the edge-chromatic number of planar regular multi-graphs. From an algorithm point of view, a proof of the dichotomy conjecture of Feder and Vardi is recently given by characterizing digraphs whose corresponding homomorphism problem is polynomial time. The extension of this work to signed digraphs would be a result of high interest.

The PARHéRo project explicitly aims to increment synergies between scientific and industrial research to anticipate and control the evolution of heterogeneous robotic platforms in complex, unknown and/or hostile environments. The successful completion of robotic missions is ensured by the high degree of autonomy of the platforms, a central element for their robustness, which is achieved through learning, planning and the supervision of the execution of intelligent behaviour. The project aims firstly at providing autonomous multi-robot platforms with a mission specification language able to express both the objectives and the requirements on the state of the robots and the characteristics of the planning model. In order to test the specification language, the project will aim to generate case studies that are coherent with the applications envisaged for future defence and security systems. The mission specification language is also the means by which the results of the learning, achieved by each of its elements, can be shared in the fleet of heterogeneous robots. Autonomous decision making, or even interactive planning with a human overseeing the mission, strives for platform resilience to unexpected, dangerous, or unpredictable events in the environment. In this context, the use of domain knowledge -- whether prior or on-the-fly during the mission execution phase -- can provide a quicker and better solution to the problems faced. This hybridisation between Automatic Planning in Artificial Intelligence and Machine Learning ensures the robustness, adaptability, and resilience of the fleet of heterogeneous robots, all participating in the same strategic objective. Automated planning and machine learning (in particular Reinforcement Learning) are characterized by complementary views on decision making: the former relies on previous knowledge used to create a model and computation of a solution from this model, while the latter on interaction with the world, and repeated experience. Reinforcement learning, can start without any previous knowledge, and allows robots to robustly adapt to the environment, but often necessitates an infeasible amount of experience. Planning allows robots to carry out different tasks in the same domain, without the need to acquire additional knowledge about the domain or about each one of them, but relies strongly on the accuracy of the model. Furthermore, the search space of a planner with partial knowledge about the environment can grow exponentially in the number of possible states, making the planning process practically unfeasible. However, even a small injection of knowledge from prior learning about the model can greatly improve the performance of the solution search. This a priori knowledge, which can come from learning phases of intelligent behaviour, allows the refinement of meta-heuristics, macro-actions, or even hierarchical task decompositions. Learning techniques can also be used to improve the decisions made by a group of robots, such as mission optimisation against opposing criteria. Whether they are high-level strategies or purely reactive components, the coordination of a fleet of autonomous mobile robots requires the transmission of information learned by each robot based on local information, provided that the robustness conditions of the communication network are maintained. Otherwise, the estimation by each robot of the global situation is necessary to guarantee the autonomy of the robot fleet, and the robustness of the mission.

This project will develop formal methods for the verification of distributed systems with unbounded numbers of participants, where the communication network changes at runtime. In addition to classical properties of parallel systems, such as deadlock-freedom and absence of critical section violations, we consider convergence and self-stabilisation. The project is structured in three axes. The first axis concerns the development of appropriate models of distributed systems, by capturing the essential aspects of reconfiguration and mobility in modern computer systems. We define models based on combinations of logic with automata and game theory, with stochastic aspects, thus enabling the study of decidability and complexity classes for the verification problems considered. The second axis concentrates on finding (semi-)algorithms capable of analyzing some classes of distributed systems. The third axis is about the evaluation of our methods, by applying the developed (semi-)algorithms to verify correctness of well-known distributed algorithms and protocols (leader election, consensus, topological coverage, routing, etc.). These distributed algorithms are currently a challenge for existing software verification methods, that deal mostly with shared-memory multi-threading programs.

Magnetic resonance (MR) imaging plays a crucial role in the detection of pathologies, the study of brain organization and the clinical research. Every day, a vast amount of data is produced in clinical settings and this number is increasing rapidly, which prevents the use of manual approaches for data analysis. As a result, the development of reliable segmentation techniques for the automatic extraction of anatomical structures is becoming an important field of quantitative MR analysis. In the DeepVolBrain project, the final goal is to develop a new generation of quantitative MRI analysis methods to cope with the rise of BigData in neuroimaging and, ultimately, to generate new knowledge. Moreover, the proposed methods will be implemented in open access to the entire community through a web platform. Objective A: First, we propose to develop novel methods by addressing the current limitations of Deep Learning (DL) in neuroimaging. DL is a fast-growing field in computer vision that has recently obtained many successes. However, so far, results obtained by DL for MRI segmentation are not as good as expected. The limited performance of DL in neuroimaging seems resulting from several factors such as few training data or large memory requirement. First, we propose to address the problem related to the limited number of training data by increasing the size of training library and by reducing the number of required training images. To this end, we will develop new data augmentation strategy and innovative DL architectures enable to improve learning speed and to reduce the number of required training images. Second, to address the memory issue related to DL, we will propose ensemble learning strategy based on locally adaptive 3D CNN. Finally, the last factor limiting the performance of DL is the quality of preprocessing to compensate for the image heterogeneity. Based on our extensive expertise, we will integrate the developed DL segmentation methods into robust pipelines. Objective B: The emergence of very large datasets opens up new challenges related to BigData as defined by the usual 3Vs model (Volume, Variety and Velocity). The fast and robust pipelines developed for Objective A will address these challenges by proposing new tools able to process large Volume of data, to cope with image Variety from different datasets and to propose high Velocity thanks to GPU-based computing. However, two Vs have been recently added to the usual 3Vs Big Data model – Veracity and Value. In medical imaging, the reliability is related to the questions of quality control (QC) and traceability. Therefore, to ensure Veracity of the produced results, we need to propose advanced QC and to estimate a confidence of the produced results. In the DeepVolBrain project, automatic QC and error correction will be integrated into pipelines to increase the confidence in the results produced. Moreover, an extensive validation over large scale datasets will be carried out. Finally, the proposed tools will be applied to large datasets including pathological cases to demonstrate the Value and the capability of the proposed project to produce new knowledge. Objective C: Finally, we will make our tools freely available by deploying a web platform. In the past, with the volBrain platform, we developed an original platform in a fully open access hosted at Valencia in Spain. This platform already processed more than 75 000 MRI in 3 years. This unexpected very high number of online processing pushes us to investigate new strategies to make our platform more scalable and to ensure its sustainability. In this project, we will first achieve the deployment of a second site in France at the LaBRI. Second, we will propose a new scalable and flexible architecture. This new architecture will enforce the security and privacy of the data. Finally, each of developed tools will be integrated in this new platform.