Search

We propose a novel learning-based formulation for visual localization of vehicles that can operate in real-time in city-scale environments. Visual localization algorithms determine the position and orientation from which an image has been captured, using a set of geo-referenced images or a 3D scene representation. Our new localization paradigm, named Implicit Pose Encoding (ImPosing), embeds images and camera poses into a common latent representation with 2 separate neural networks, such that we can compute a similarity score for each image-pose pair. By evaluating candidates through the latent space in a hierarchical manner, the camera position and orientation are not directly regressed but incrementally refined. Very large environments force competitors to store gigabytes of map data, whereas our method is very compact independently of the reference database size. In this paper, we describe how to effectively optimize our learned modules, how to combine them to achieve real-time localization, and demonstrate results on diverse large scale scenarios that significantly outperform prior work in accuracy and computational efficiency., Accepted at WACV 2023

International audience; In this paper we propose RRDV, a system for robot-to-robot encounter detection. We use low-cost ultrasound sensor and time-synchronized mobile robots to detect when two robots are facing one another. Ultrasound ranging is triggered by the control application on a computer. The application sends a ranging command to the gateway, which broadcasts it to the mobile robots over the radio. Robots synchronize their ultrasound trigger pin with the start of frame event and send back the notifications with measured distances using Time-Division Multiple Access (TDMA). The system then finds the encounters by searching for timestamps where the difference in distance reported by two robots is less then 1 cm. In the current implementation, the system achieves a 20 Hz distance measurement update rate. RRDV is validated experimentally using 5 mobile robots which are controlled by the users and moved randomly. We implemented a Computer Vision (CV) algorithm for tracking mobile robots as they move and detect when they are facing one another. The CV algorithm is used as the ground truth for the experimental evaluation. The results show 96.7% successfully detected robot encounters, when the duration of the encounter is more than 5 s.

International audience; The study of human-human interactions is essential for a better understanding of human behaviour during collaborative tasks. This knowledge is not only interesting in life science but can also be useful in robotic science. Indeed, to efficiently assist a human partner during a human-robot collaboration, the robot needs to be as reactive as a human would be. This can only be achieved by embedding a model of human behaviour into the robot control scheme. In this paper, a human-humanoid robot collaboration to carry a table is tackled. First, the experimental Center of Mass (CoM) trajectories of a table carried by 20 pairs of subjects to various goal positions are studied and modeled using an optimal control problem. Then, based on this model, a prediction process which accurately predicts the table trajectories is designed. Finally, this prediction process is coupled with the robot Walking Pattern Generator (WPG). Using a torque whole-body controller, this framework is tested in simulation on Gazebo on a TALOS humanoid robot model. In this simulation, the robot actively assists a simulated human partner in lifting and carrying a table to an unknown goal position.

International audience; Quadruped robots have proved their robustness to cross complex terrain despite little environment knowledge. Yet advanced locomotion controllers are expected to take advantage of exteroceptive information. This paper presents a complete method to plan and control the locomotion of quadruped robots when 3D information about the surrounding obstacles is available, based on several stages of decision. We first propose a contact planner formulated as a mixed-integer program, optimized on-line at each new robot step. It selects a surface from a set of convex surfaces describing the environment for the next footsteps while ensuring kinematic constraints. We then propose to optimize the exact contact location and the feet trajectories at control frequency to avoid obstacles, thanks to an efficient formulation of quadratic programs optimizing Bezier curves. By relying on the locomotion controller of our quadruped robot Solo, we finally implement the complete method, provided as an open-source package. Its efficiency is asserted by statistical evaluation of the importance of each component in simulation, while the overall performances are demonstrated on various scenarios with the real robot.

Mapping a structure, such as a metal plate in a ship hull, using acoustic echoes typically requires making assumptions on its shape (e.g. rectangular). This work introduces a more general beamforming approach to recover the geometry of arbitrary polygonal-shaped bounded areas from acoustic reflections. Our method only requires a single omnidirectional emitterreceiver acoustic device mounted on a mobile platform. We apply beamforming to the acoustic measurements in the geometrical boundary space. We subsequently retrieve the edges from the beamforming results via the minimization of a regularized cost criterion, using a simulated annealing optimizer. We also design a boundary rejection criterion so that their exact number can be recovered based only on a specified upper bound. We assess our method on different geometries in a simulation environment and a real-world setting using ultrasonic guided waves measurements. The results demonstrate that it is efficient for achieving the targeted objectives.

International audience; This paper proposes a decentralized and online optimal perception-aware strategy for multi-robot systems. The aim is to maximize the information collected along the planned trajectory about the relative configurations of the robots and, hence, to minimize the localization uncertainty. This is done by leveraging the so-called Constructability Gramian (CG), which can quantify the information about the future state of a nonlinear system. We show that, thanks to a proper change of coordinates, the CG can be computed in a decentralized way with only minor approximations. This allows for formulating an online and decentralized trajectory generation problem for optimal localization. To show the effectiveness of the approach, we consider as case study the localization of a quadrotor group with noisy distance measurements and sensing constraints. The results show the interest of the proposed approach.

We introduce RoboPose, a method to estimate the joint angles and the 6D camera-to-robot pose of a known articulated robot from a single RGB image. This is an important problem to grant mobile and itinerant autonomous systems the ability to interact with other robots using only visual information in non-instrumented environments, especially in the context of collaborative robotics. It is also challenging because robots have many degrees of freedom and an infinite space of possible configurations that often result in self-occlusions and depth ambiguities when imaged by a single camera. The contributions of this work are three-fold. First, we introduce a new render & compare approach for estimating the 6D pose and joint angles of an articulated robot that can be trained from synthetic data, generalizes to new unseen robot configurations at test time, and can be applied to a variety of robots. Second, we experimentally demonstrate the importance of the robot parametrization for the iterative pose updates and design a parametrization strategy that is independent of the robot structure. Finally, we show experimental results on existing benchmark datasets for four different robots and demonstrate that our method significantly outperforms the state of the art. Code and pre-trained models are available on the project webpage https://www.di.ens.fr/willow/research/robopose/., Accepted at CVPR 2021 (Oral)

International audience; In this paper, a decentralized autonomous controller aimed to control a fleet of quadrotors is designed, based on the iterative generation and exploitation of logged traces. The presented approach, inspired by model predictive control, aims to maintain the geometrical configuration for a set of quadrotors led by remotely controlled leaders. The novelty of this approach is to rely on inexpensive commercial off-the-shelf sensors (as opposed to positioning systems and/or cameras) that only measure the distance among quadrotors. In the first phase (trace generation) quadrotors are operated using randomized controllers based on domain knowledge, and their trajectories are registered. In the exploitation phase, a policy is learned from the traces generated in the previous phase, and the policy is iteratively refined, to achieve a robust reactive control of each quadrotor agent. Extensive experiments using RotorS, a Software In the Loop (SITL) framework in Gazebo simulator demonstrates the efficiency of the approach, and its ability to preserve the flocking structure of the quadrotors, following the (remotely and independently controlled) leaders.

International audience; Vehicle-to-vehicle (V2V) messaging is an indispensable component of connected autonomous vehicle systems. Although V2V standards have been specified by the European Union, United States, and Japan, the deployment phase represents mixed traffic in which connected and legacy vehicles co-exist. To enhance cooperative awareness in this mixed traffic, we assessed the special roadside unit that we developed in our previous work that generates required V2V messages on behalf of sensed target vehicles. In this paper, we extend our earlier work to propose a system called Grid Proxy Cooperative Awareness Message to broaden the cooperative awareness message dissemination area by connecting infrastructure using high-speed roadside networks. To minimize delay in message delivery, we designed the proposed system to use edge computing. The proposed scheme delivers cooperative messages to a wider area with a low delay and a high packet delivery ratio by prioritizing packets by their respective safety contributions. Our simulation results indicate that the proposed scheme efficiently delivers messages in heavy road traffic conditions modeled on real maps of Tokyo and Paris.

International audience; Connected components labeling and analysis for dense images have been extensively studied on a wide range of architectures. Some applications, like particles detectors in High Energy Physics, need to analyse many small and sparse images at high throughput. Because they process all pixels of the image, classic algorithms for dense images are inefficient on sparse data. We address this inefficiency by introducing a new algorithm specifically designed for sparse images. We show that we can further improve this sparse algorithm by specializing it for the data input format, avoiding a decoding step and processing multiple pixels at once. A benchmark on Intel and AMD CPUs shows that the algorithm is from ×1.6 to ×2.5 faster on sparse images.

International audience; Accurate detection of objects in 3D point clouds is a central problem for autonomous navigation. Most existing methods use techniques of hand-crafted features representation or multi-sensor approaches prone to sensor failure. Approaches like PointNet that directly operate on sparse point data have shown good accuracy in the classification of single 3D objects. However, LiDAR sensors on Autonomous Vehicles generate a large scale point cloud. Real-time object detection in such a cluttered environment still remains a challenge. In this study, we propose Attentional Point- Net, which is a novel end-to-end trainable deep architecture for object detection in point clouds. We extend the theory of visual attention mechanisms to 3D point clouds and introduce a new recurrent 3D Localization Network module. Rather than processing the whole point cloud, the network learns where to look (finding regions of interest), which significantly reduces the number of points to be processed and inference time. Evaluation on KITTI car detection benchmarkshows that our Attentional PointNet achieves comparable results with the state-of-the-art LiDAR-based 3D detection methods in detection and speed.

Path tracking controllers for an autonomous vehicle are often designed by using either a dynamic model or a kinematic one and some models are related to wheel-ground contact, that makes the efficiency of the controller highly dependent on the ground parameters estimation, especially for off-road mobile robots intended to navigate in open environments. This paper proposes a new nonlinear observer designed to estimate the front and rear contact cornering stiffnesses in real time, that are related both on tire and soil proprieties. The latter is estimated using steering angles as well as yaw rate and lateral velocity, which are provided by a preliminary Kalman-Bucy observer. The performance of the proposed nonlinear observer combined with the LQR controller is evaluated by both advanced simulations and experiments in real conditions at different speeds.

International audience; The ability to predict the future intended movement is crucial for collaborative robots to anticipate the human actions and for assistive technologies to alert if a particular movement is non-ergonomic and potentially dangerous for the human health. In this paper, we address the problem of predicting the future human whole-body movements given early observations. We propose to predict the continuation of the high-dimensional trajectories mapped into a reduced latent space, using autoencoders (AE). The prediction is based on a probabilistic description of the movement primitives (ProMPs) in the latent space, which notably reduces the computational time for the prediction to occur, and hence enables to use the method in real-time applications. We evaluate our method, named AE-ProMPs, for predicting future movements belonging to a dataset of 7 different actions performed by a human, recorded by a wearable motion tracking suit.

International audience; We present here a framework to integrate into a motion planning method the interaction zones of a moving robot with its future surroundings, the reachable interaction sets. It can handle highly dynamic scenarios when combined with path planning methods optimized for quasi-static environments. As a demonstrator, it is integrated here with an artificial potential field reactive method and with a Bézier curve path planning. Experimental evaluations show that this approach significantly improves dynamic path planning methods, especially when the speeds of the obstacles are higher than the one of the robot. The presented approach is used together with a global planning approach in order to handle complex static environments in presence of fast-moving obstacles. When the ego vehicle is not holonomic the presented approach is able to take dynamic constraints into account, which improve the prediction accuracy.

International audience; This article introduces an open source tool for simulating autonomous vehicles in complex, high traffic, scenarios. The proposed approach consists on creating an hybrid simulation, which fully integrates and synchronizes two well known simulators: a microscopic, multi-modal traffic simulator and a complex 3D simulator. The presented software tool allows to simulate an autonomous vehicle, including all its dynamics, sensors and control layers, in a scenario with a very high volume of traffic. The hybrid simulation creates a bi-directional integration, meaning that, in the 3D simulator, the ego-vehicle sees and interacts with the rest of the vehicles, and at the same time, in the traffic simulator, all additional vehicles detect and react to the actions of the ego-vehicle. Two interfaces, one for each simulator, where created to achieve the integration, they ensure the synchronization of the scenario, the state of all vehicles including the ego-vehicle, and the time. The capabilities of the hybrid simulation was tested with different models for the ego-vehicle and almost 300 additional vehicles in a complex merge scenario.

International audience; Synchronization is an indissociable part of social interactions between humans, especially in gestural communication. With the emergence of social robotics and assistance robots, it becomes paramount for robots to be socially accepted and for humans to be able to connect with them. As a consequence, synchronization mechanisms should be inherent to any robot controllers, allowing the adaption to the interacting partner in any rhythmic way necessary. In this paper, plastic Central Pattern Generators (CPG) have been implemented in the joints of the robot Pepper that has to learn to wave back at a human partner. Results show that the CPG-based controller leads to adaptive waving synchronized with the human partner, thus proving that the CPG-based controller can achieve synchronization.

International audience; A key factor of a successful Human-Robot Interaction (HRI) is the robot's ability to understand the messages transmitted by a human user, especially their nonverbal and paraverbal components. Our work focuses on the robot's ability to detect if a person is deceitful or honest, based on an evaluation of their physiological state and using only noninvasive techniques, such as thermal and RGB-D imaging. In particular, we currently study and evaluate several physiological manifestations during an experiment consisting of an interrogation conducted by a robot that follows a mock-up crime, then correlate them with the truth values of the participants' answers. The first results are promising.

International audience; — In the context of a worldwide aging population, it is important to find solutions to help the elderly maintain their cognitive functions. This research was done in the context of the ENRICHME. We investigate the outcome of a 5 day intensive cognitive stimulation with an elderly individual. Each day was composed of two sessions (one at 11am, and one at 3pm). During each session the participant played three cognitive games (i.e., digit cancellation, integer matrix task, Stroop game), two of them having two difficulty levels. The mood of the participant was also recorded before and after each interaction session. Evidence was found that even after a few sessions, the performance of the participant increased for all games. The performance of each game and each difficulty level was analyzed based on the interaction time (11am or 3pm) and the interaction day. A detailed analysis of the performance is presented together with a discussion of these results.

We propose a kinematic motion strategy which allows performing a sequential merging of a set of curvature-constrained wheeled vehicles in order to obtain a platoon formation - one of the basic motion tasks defined for intelligent autonomous vehicles. The proposed design methodology is motivated by the lining-up maneuvering characterizing the N-trailer structures. The main emphasis has been paid on the safety guarantees of the dynamic navigation of the fleet of vehicles. In this context, we provide formal conditions under which all the vehicles during the maneuvers keep their motion in a prescribed road width, their longitudinal velocities and motion curvatures are constrained to the prescribed bounded ranges, and the minimal safe inter-vehicle distances are not violated. Formal considerations have been complemented by exemplary results of numerical simulations for a set of five vehicles.

With large-scale penetration of renewable energy generators, the prediction uncertainty of power demand has highly increased in recent years. In this paper, we propose a new approach for economic dispatch to obtain the tolerance interval of electric facilities' operations, which minimizes the worst-case dispatch cost while guaranteeing the adaptivity of actual dispatch schedule to any realization of net power demand predicted. Representing the dispatch cost as a quadratic function of the power outputs by fuel-based generators and the charginydischarging power of batteries, we formulate this problem as a min-max-min robust optimization problem. To solve it, we first show that the worst-case dispatch schedule consists of either maximum or minimum values in the tolerance intervals at each time slot. Then, we reduce the problem to a minimax problem and solve it using Benders decomposition to deal with the high computational cost. Finally, we verify the effectiveness of our proposed method by numerical simulations.

Creating successful human-robot collaboration requires robots to have high-level cognitive functions that could allow them to understand human language and actions in space. To meet this target, an elusive challenge that we address in this paper is to understand object-directed actions through grounding language based on visual cues representing the dynamics of human actions on objects, object characteristics (color and geometry), and spatial relationships between objects in a tabletop scene. The proposed probabilistic framework investigates unsupervised Part-of-Speech (POS) tagging to determine syntactic categories of words so as to infer grammatical structure of language. The dynamics of object-directed actions are characterized through the locations of the human arm joints - modeled on a Hidden Markov Model (HMM) - while manipulating objects, in addition to those of objects represented in 3D point clouds. These corresponding point clouds to segmented objects encode geometric features and spatial semantics of referents and landmarks in the environment. The proposed Bayesian learning model is successfully evaluated through interaction experiments between a human user and Toyota HSR robot in space.

This work proposes an optimal trajectory generation and a controller design for a mobile robot. The objective consists of creating a new performance guidance module for the mobile robot in a closed environment. The reference trajectory can be calculated by solving an optimization problem. The properties of the differential flatness, the B-spline and the direct collocation method are exploited to convert the constrained optimization problem into a nonlinear programming one that can be easily resolved by a classic solver. After that, the obtained optimal reference trajectory is used in the kinematic mobile robot model. A flatness-based controller is designed to ensure the trajectory tracking. The numerical simulations are done in order to evaluate the proposed optimal trajectory generation approach and the tracking strategy.

International audience; In the context of scene modelling, understanding, and landmark-based robot navigation, the knowledge of static scene parts and moving objects with their motion behaviours plays a vital role. We present a complete framework to detect and extract the moving objects to reconstruct a high quality static map. For a moving 3D camera setup, we propose a novel 3D Flow Field Analysis approach which accurately detects the moving objects using only 3D point cloud information. Further, we introduce a Sparse Flow Clustering approach to effectively and robustly group the motion flow vectors. Experiments show that the proposed Flow Field Analysis algorithm and Sparse Flow Clustering approach are highly effective for motion detection and seg-mentation, and yield high quality reconstructed static maps as well as rigidly moving objects of real-world scenarios.

International audience; With the advent of low-cost RGBD sensors, many solutions have been proposed for extraction and fusion of colour and depth information. In this paper, we propose new different fusion approaches of these multimodal sources for people detection. We are especially concerned by a scenario where a robot evolves in a changing environment. We extend the use of the Faster RCNN framework proposed by Girshick et al. [1] to this use case (i), we significantly improve performances on people detection on the InOutDoor RGBD People dataset [2] and the RGBD people dataset [3] (ii), we show these fusion handle efficiently sensor defect like complete lost of a modality (iii). Furthermore we propose a new dataset for people detection in difficult conditions: ONERA.ROOM (iv).

International audience; — This paper presents the analytic solution of a fundamental open problem in the framework of state esti-mation/nonlinear observability, which is the Unknown Input Observability problem (UIO problem). The problem consists in deriving the analytic criterion that allows us to automatically obtain the state observability in presence of disturbances (or unknown inputs). In other words, the problem is to extend the well known observability rank condition to the case when the dynamics are also driven by unknown inputs. Enunciated in the seventies by the control theory community, this problem was only solved in the linear case. The solution here provided holds for nonlinear systems in presence of a single unknown input. The first part of the paper presents this analytic solution. Very surprisingly, the complexity of the overall analytic criterion is comparable to the complexity of the observability rank condition. The second part of the paper applies this analytic criterion to a robotics system when its dynamics are affected by an external disturbance (e.g., due to the presence of wind). To corroborate the results of our observability analysis we perform extensive simulations and we show that, a simple estimator based on an Extended Kalman Filter, provides results that agree with what we could expect from the observability analysis.

International audience; The knowledge of the static scene parts and the moving objects in a dynamic scene plays a vital role for scene modelling, understanding, and landmark-based robot navigation. The key information for these tasks lies on semantic labels of the scene parts and the motion trajectories of the dynamic objects. In this work, we propose a method that segments the 3D feature trajectories based on their motion behaviours, and assigns them semantic labels using 2D-to-3D label transfer. These feature trajectories are constructed by using the proposed trajectory recovery algorithm which takes the loss of feature tracking into account. We introduce a complete framework for static-map and dynamic objects' reconstruction, as well as semantic scene understanding for a calibrated and moving 2D-3D camera setup. Our motion segmentation approach is faster by two orders of magnitude, while performing better than the state-of-the-art 3D motion segmentation methods, and successfully handles the previously discarded incomplete trajectory scenarios.

International audience; Modelling nonverbal communication in robotics is a crucial issue to improve Human Robot interactions (HRI). Among several nonverbal behaviours we focus in this article on unintentional rhythmic entrainment and synchronization which has been proven to be highly important in intuitive and natural Human Human communication. Hence, the rising question is whether or no this phenomenon can be reproduced in a context of HRI and what are the prerequisites to ensure its emergence. In this paper, we study rhythmical interactions during imitation games between a NAO robot and naive subjects. We analysed two main types of interactions, a first where NAO performs movements at a fixed rhythm (unidi-rectional) and a second one where the robot is able to adopt the human motion dynamic (bidirectional) using a neural modelling of the entrainment effect based on dynamical systems. We show that using such model allows us to reach synchronization during the interactions and that both partners (robot and human) adapt their frequency as observed in natural HHI. This puts forward the importance of bidirec-tionality for HRI. Moreover, the participants shifted their motion dynamics during the interaction without noticing it, proving the presence of such unintentional rhythmic entrainment in HRI.

International audience; Building and maintaining the sustainability of European projects requires advanced communication and team investment, shared working methods, common decisions and symbiosis in key items such as selection requirement and process, terms of mobility between consortium members or teaching content. It requires jointness in various perspectives, aspects and scales, with their own conditions and challenges. This abstract describes this prerequisite of jointness through the local campus of Le Creusot (University of Burgundy) for our International Programme in Computer Vision and Robotics (VIBOT). The main challenge can be summarized in one single question: How to maintain and improve our level of jointness beyond the European framework of an Erasmus Mundus label?

International audience; The localization of a vehicle is a central task of autonomous driving. Most of the time, it is solved by considering a single algorithm with a few sensors. In this paper, we propose a cooperative fusion architecture based on two main algorithms: a laser-based Simultaneous Localization And Mapping (SLAM) process and a lane detection and tracking approach using a single camera. Both algorithms are designed individually as cooperative fusion processes where other sensors (GPS and proprioceptive information) and dedicated maps are integrated to strengthen the advantages of each system. The whole architecture is formalized around key components (ego-vehicle, roadway, obstacle and environment). A final decision layer, that takes into account the state of each algorithm, allows the system to choose the most appropriate ego-vehicle localization mean based on the current road situation and the environmental context.

International audience; Tactile sensing feedback provides feasible solutions to robotic dexterous manipulation tasks. In this paper, we present a novel tactile-based framework for detecting/correcting slips and regulating grasping forces while manipulating de-formable objects with the dynamic center of mass. This framework consists of a tangential force based slip detection method and a deformation prevention approach relying on weight estimation. Moreover, we propose a new strategy for manipulating deformable heavy objects. Objects with different stiffnesses, surface textures, and centers of mass are tested in experiments. Results show that proposed approaches are capable of handling objects with uncertainties in their characteristics, and also robust to external disturbances.

International audience; Developing autonomous vehicles capable of dealing with complex and dynamic unstructured environments over large-scale distances, remains a challenging goal. One of the major difficulties in this objective is the precise localization of the vehicle within its environment so that autonomous navigation techniques can be employed. In this context, this paper presents a methodology to map building and to efficient pose computation which is specially adapted for cases of large displacements. Our method uses hybrid robust RGB-D cost functions that have different convergence properties, whilst exploiting the visibility rotation invariance given by panoramic spherical images. The proposed registration model is composed of a RGB and point-to-plane ICP cost in a multi-resolution framework. We close up the paper presenting mapping and localization results in real outdoor scenes.

International audience; Cooperative Intelligent Transportation Systems (CITS) are being developed to make road traffic safer and more efficient. Vehicle to vehicle (V2V) communication plays an important role for the cooperative awareness of vehicles in all the CITS standards in EU, US, and Japan. Essentially the CITS application relies on the exchanged data by V2V messages (called Cooperative Awareness Message (CAM) in EU). However, the V2V communication suffers from two significant problems: 1) mixed environment and 2) obstacle interference. First, the CITS application cannot be aware of a non-CAM-aware vehicle because of low market penetration ratio of the V2V message transmitter in the early deployment phase. Second, the V2V messages may not reach the receivers because of interference by the obstacle between the vehicles. To solve these problems, the proposed system called Proxy CAM generates the necessary V2V messages in the roadside unit on behalf of sender vehicles (or non-CAM-aware vehicle) by sensing the vehicle. We design the system based on the ITS Station architecture standardized in ISO/ETSI, working with any vehicle sensing technology. Moreover, we implement the prototype system with a stereo vision for the vehicle sensing. The prototype system is evaluated in an indoor test and also validated in a field test in Tokyo with some modifications in the link layer to adapt the Japanese regulation. The results show that Proxy CAM significantly improves the coverage of V2V messaging while the system overhead is limited.

Solving the direct kinematics (DK) of parallel robots, i.e. finding all the possible poses of the platform for given input values, is most of the time a difficult problem. This is evidently true for cable-driven parallel robots (CDPR) that are more complex than classical parallel robot because of the unilateral nature of the cable actions but also if cable deformations are taken into account. Furthermore there are many different deformable cable models and developing a DK algorithm for each of them will be a tremendous task. Consequently using a numerical continuation scheme that starts from the DK solutions for non-deformable cables and then moves toward the solution for deformable cables appears to be an interesting approach. To use this scheme we have to assume that the cable model relies on physical parameters that have limit values for which the deformable cable acts like a non-deformable one. Under that assumption we are then able to derive a possible maximum of solutions for the DK, whatever is the cable model. We then apply this approach for a specific complex cable model, the catenary case, to show that this approach is computer efficient but requires to address difficult theoretical issues in order to obtain a solving algorithm that is guaranteed to determine all DK solutions.

Unmanned Aerial Vehicles (UAVs) fleets are becoming more apparent in both military and civilian applications. However security of these systems still remains unsatisfactory if a strong adversary model with a high attack potential (i.e. the adversary has capabilities and knowledge to capture a UAV, to perform side-channel or fault injection or other physical, software or combined attacks in order to gain access to some secret data like cryptographic keys, mission plan, etc.) is considered. The aim of this position paper is to draw security requirements for this kind of adversaries and to propose theoretical solutions based on an embedded Secure Element (SE) that could help to accommodate these requirements. Finally, our proposal on how to use these SEs to secure Autonomous UAVs fleets is presented.

International audience; This paper considers collaborative stereo-vision as a mean of localization for a fleet of micro-air vehicles (MAV) equipped with monocular cameras, inertial measurement units and sonar sensors. A sensor fusion scheme using an extended Kalman filter is designed to estimate the positions and orienta-tions of all the vehicles from these distributed measurements. The estimation is completed by a formation control to maximize the overlapping fields of view of the vehicles. Experimental tests for the complete perception and control loop have been performed on multiple MAVs with centralized processing on a ROS ground station.

International audience; —We introduce a novel rehabilitation tool to treat Post-Fall Syndrome (PFS) in older adults: a serious game, called Virtual Promenade, combined with a haptic chair imitating the hips movements of human walk. We report on the user-centered design of our prototype, following " living lab " principles, which was well received by our test participants. This system aims at addressing the psycho-motor consequences of older adults' falls; they are often neglected in current post-fall care practices. We first checked for feasibility and tolerability of such interventions. We then applied a living lab participatory design approach, involving health care professionals and older adults, to build the Virtual Promenade prototype. We found that patients with PFS tolerated the system well and that there were no major obstacles to feasibility. We also report that the aesthetics of the virtual environment is an important motivational factor for older adults and discuss our results in searching for the most suitable game controller for this type of patients and game. Finally, we observed that the chairs' movements improved the immersion in the game.

International audience; As an important movement of the daily living activities, sit-to-stand (STS) movement is usually a difficult task facing elderly and dependent people. To provide appropriate power assistance for the sit-to-stand movement, a novel intention-based Active Impedance Control (AIC) strategy applied on a lower limb exoskeleton is proposed in this paper. The AIC is able to adapt the mechanical impedance of the human-exoskeleton system towards a desired one using the exoskeleton's power assistance. In the AIC structure, a human joint torque observer is designed to estimate the human joint torques using joint angles information instead of electromyography (EMG) or force/torque sensors; a time-varying desired impedance model is proposed according the wearer's lower limb motion ability. Simulations were implemented to illustrate the characteristics and performances of the proposed approach. Experiments with a healthy subject were carried out to evaluate the effectiveness of the proposed method. The experiments show satisfactory results in terms of appropriate power assist based on the wearer's motion intention.

International audience; The wide acceptance of wheelchairs in the society may be associated to their simplicity, and to the important mobility ease they provide to people with disability. Nonetheless, users with extreme mobility impairments still experience some cognitive and physical steering challenges that limit their capacity to navigate even simple environments. Assistive wheelchairs with functionalities to enable customisation of user handicap are therefore still required. The diversity of user disabilities and needs, and the corresponding assistive measures, necessitate proper understanding of the user. Modelling and integration of user driving behaviour into the control loop in order to monitor user inputs, and execute necessary corrective steps in troublesome situations, may just be one way of providing the assistance that corresponds to needs of the user. This paper attempts to identify the steering behaviour of an individual wheelchair user in a normal living room environment. The preliminary navigation experiments required to obtain the steering data are performed in a virtual world on an augmented wheelchair platform, and the participant is observed to exhibit expected navigational behaviours. Least square method is considered in the estimation of model parameters for the non-linear steering model with interesting regression analysis outcome.

International audience; — This paper presents a distributed saturated formation control for a multiple quadrotors (UAVs) system with leader-follower structure. Each UAV has local and limited neighbors. Stability analysis for the multi-UAV system with fixed and switching topology is given, using the theories of convex hull and perturbed matrix. Simulation results show that the quadrotors with the proposed strategy can achieve consensus in presence of changeable leaders with switching topology. The real-time experiment shows that the quadrotors can track the desired formation trajectory by using the proposed formation strategy.

International audience; Members of a team are able to coordinate their actions by anticipating the intentions of others. Achieving such implicit coordination between humans and robots requires humans to be able to quickly and robustly predict the robot's intentions, i.e. the robot should demonstrate a behavior that is legible. Whereas previous work has sought to explicitly optimize the legibility of behavior, we investigate legibility as a property that arises automatically from general requirements on the efficiency and robustness of joint human-robot task completion. We do so by optimizing fast and successful completion of joint human-robot tasks through policy improvement with stochastic optimization. Two experiments with human subjects show that robots are able to adapt their behavior so that humans become better at predicting the robot's intentions early on, which leads to faster and more robust overall task completion.

International audience; Satellite positioning is a key technology for autonomous navigation in outdoors environments. When using standalone computation with mono-frequency receivers, positioning errors are not in accordance with the required performance. Nevertheless, since errors are strongly time-correlated, a GPS fix is quite informative if a shaping model of the positioning errors is carefully handled and made possible by exteroceptive sensors. When driving in a road with a camera detecting lane markings, the lateral error is directly observable by using a lane marking map. It can be well modeled by mixing auto-regressive and random constant models. An algebraic observability study is conducted to prove that this modeling is completely observable in a road frame. A new road invariant Extended Kalman Filter (EKF) is then presented to conserve the observability of every component of the state vector for any road whatever its orientation. The filter manages road changes by using bijective transformations that are detailed. Real experimental results indicate that the performance of the estimation process is significantly improved compared to a classic EKF that is implemented in a fixed working frame with a less informative error modeling.

International audience; — Mapping evolving environments requires an update mechanism to efficiently deal with dynamic objects. In this context, we propose a new approach to update maps pertaining to large-scale dynamic environments with semantics. While previous works mainly rely on large amount of observations, the proposed framework is able to build a stable representation with only two observations of the environment. To do this, scene understanding is used to detect dynamic objects and to recover the labels of the occluded parts of the scene through an inference process which takes into account both spatial context and a class occlusion model. Our method was evaluated on a database acquired at two different times with an interval of three years in a large dynamic outdoor environment. The results point out the ability to retrieve the hidden classes with a precision score of 0.98. The performances in term of localisation are also improved.

A common goal of robot control architecture designers is to create systems that are sufficiently generic to be adapted to different robot hardware. Beyond code re-use from a software engineering standpoint, having a common architecture could lead to long-term experiments spanning multiple robots and research groups. This paper presents a first step toward this goal with HBBA, a Hybrid Behavior-Based Architecture first developed on the IRL-1 humanoid robot and integrating an Adaptive Resonance Theory-based episodic memory (EM-ART). This paper presents the first step of the adaptation of this architecture to two different robots, a Meka M-1 and a NAO from Aldebaran, with a simple scenario involving learning and sharing objects' information between both robots. The experiment shows that episodes recorded as sequences of people and objects presented to one robot can be recalled in the future on either robot, enabling event anticipation and sharing of past experiences.

Safe Autonomous Vehicles (AVs) will emerge when comprehensive perception systems will be successfully integrated into vehicles. Advanced perception algorithms based on occupancy grids were developed for AV prototypes. These algorithms estimate the position and speed of every obstacle in the environment by using data fusion from multiple sensors. Computational requirements of such fusion prevent their integration into AVs on current low-power embedded hardware. However, recent emerging many-core architectures offer opportunities to fulfill the automotive market constraints and efficiently support advanced perception applications. This paper explores the integration of the occupancy grid multi-sensor fusion algorithm into low power many-core architectures. The parallel properties of this function are used to achieve realtime performance with a power consumption less than 1W. The proposed implementation produces an occupancy grid of 500×300 cells within 6.26ms. The execution time is 6x faster than typical sensor output rates and 9x faster than previous embedded prototypes.

International audience; Home automation is now implemented in many retirement homes in order to improve elderly’s autonomy and safety. Smart homes allow to monitor the activities of elderly persons using information coming from different sensors. The ADL (Activities of Daily Living) are used to evaluate the ability of a person to perform on their own a selection of the activities which are essential for independent living in everyday life. The ADL are then used to detect deviations in a person’s behaviour. Indoor localization based on the fusion of heterogeneous data from different sensors, is then essential for ADL characteriza- tion. For this purpose, a robust data fusion method is presented in this work through a multi-modal analysis to monitor the activities of elderly people (immobility, walking, etc) in a smart home. The paper describes the installation of sensors in a Living Lab and the preliminary experimental results using a set of Pyroelectric Infra Red (PIR) sensors, Radio Frequency Identification (RFID) distance measurement and the outcome of a noise analysis. Within a set-membership framework, our algorithm for robust localization employs a multi-modal data fusion approach dealing with faulty measurements.

2014 IEEE International Conference on Robotics and Biomimetics (ROBIO), Bali, INDONESIA, DEC 05-10, 2014; International audience; The new open-source Matlab/Simulink toolbox called RT-MaG presented here generates reliable standalone robotic applications running on real-time embedded Linux targets such as tiny Computers On Module (e.g., Gumstix boards). This toolbox gives direct access from Simulink to the main communication drivers classically used in robotics: network interfaces (via UDP), asynchronous and synchronous serial port interfaces (RS232, SPI), Pulse-width-modulation (PWM), general purpose input-output (GPIO) and analog-to-digital converters (ADCs). A Simulink model is automatically converted into a standalone multi-task program, which guarantees a repeatable execution time within each sampling time. The toolbox includes efficient debug modes which detect problems such as unsuitable configurations and hardware failure. The main features of the toolbox and its structure are described here. We also discuss the real-time performances and I/Os delays and show that a control loop can be implemented at frequencies of up to 1kHz. The tests performed show that RT-MaG can be used to efficiently implement all the control laws involved in stabilizing a quadrotor.

International audience; — A sub-optimal guidance and nonlinear control scheme based on Optic Flow (OF) cues ensuring soft lunar land-ing using two minimalistic bio-inspired visual motion sensors is presented here. Unlike most previous approaches, which rely on state estimation techniques and multiple sensor fusion methods, the guidance and control strategy presented here is based on the sole knowledge of a minimum sensor suite (including OF sensors and an IMU). Two different tasks are addressed in this paper: the first one focuses on the computation of an optimal trajectory and the associated control sequences, and the second one focuses on the design and theoretical stability analysis of the closed loop using only OF and IMU measurements as feedback information. Simulations performed on a lunar landing scenario confirm the excellent performances and the robustness to initial uncertainties of the present guidance and control strategy.

Tightly coupling global positioning system (GPS) pseudorange and Doppler measurements with other sensors is a way to increase accuracy and integrity of the positioning information particularly when it is computed autonomously. Highly accurate digital maps are also more and more key components for autonomous vehicle navigation and can enhance the localization system. In this paper, a video camera is used to get relative information with respect to lane markings and dead-reckoning sensors are also integrated to provide positioning information with high availability. A reduced order state space modeling of the observation problem is proposed in order to get an efficient real-time computing system. Thanks to an adequate error model of the GPS pseudoranges and a measured input state space formalization, a Kalman filter with correlated noises is developed. Experimental results show that this tightly coupled approach clearly improves the performance in terms of accuracy and integrity compared to a loosely coupled method that uses GPS fixes computed by an external receiver.

International audience; —Recent works have focused on lane marking feature based vehicle localization using enriched maps. The localization precision of existing methods depends strongly on the accuracy of the maps which are specially customized. In this article, we propose a marking feature based vehicle localization using open source map. Our method makes use of multi-criterion confidences to infer potential errors, and in advance, to en-hance the vehicle localization. At first, the vision-based lane marking models are obtained. Meanwhile, the map-based lane markings of current state are derived from map databases. Both lane marking sources are fused together to implement vehicle localization, using a multi-kernel based algorithm. In order to further improve the localization performance, a probabilistic error model is employed to identify the possible errors. The experiments have been carried out on public database. The results show that error modeling leads to a lower average lateral error in localization result.

International audience; Most range-free techniques for indoor localization depend on the received signal strength (RSS) fingerprints. Their performances are relied to the structure of the considered indoor environments. We consider in this paper RSS-based methods: Multi-Layer Perceptron Neural Network (MLPNN), and K-nearest neighbor (KNN), and compare their performance under the same indoor environment. One of the advantages focused by the choice of these techniques is their robustness against external disturbances that may affect the received RSS signal. Moreover, we propose a new metric to enhance the performance of the KNN method, called d-nearest neighbor. In order to test the different techniques, we build a heterogeneous fingerprint database with different resolutions. The obtained results show the efficiency of the proposed enhancement in the case of a heterogeneous high resolution database.