ParisTech

About: ParisTech is a education organization based out in Paris, France. It is known for research contribution in the topics: Finite element method & Residual stress. The organization has 1888 authors who have published 1965 publications receiving 55532 citations. The organization is also known as: Paris Institute of Technology & ParisTech Développement.

Multi-vehicle cooperative perception and augmented reality for driver assistance: A possibility to ‘see’ through front vehicle

Abstract: A typical scenario where a front vehicle (the first vehicle) occludes the view of another vehicle (the second vehicle) is often encountered in traffic environment and can be potentially dangerous. For enhancing traffic safety in this scenario, multi-vehicle cooperative perception between the two vehicles is useful. Besides, better visualization of the cooperative perception result might be needed for driver assistance. Based on these motivations, a method of multi-vehicle cooperative perception which realizes an effect of augmented reality is proposed in this paper; the effect of augmented reality here means a direct and natural visualization of the occluded environment for the second vehicle, as if a person at the second vehicle can see through the front vehicle and directly perceive the environment occluded. The proposed cooperative perception method can be used for applications of completely automated mode while the effect of augmented reality would also be convenient for driver assistance. Theoretical and technical details of the proposed method are described; field tests results are given to demonstrate the performance of the proposed method.1

Modelling the impact of climate change on sediment yield in a highly erodible Mediterranean catchment

Abstract: The assessment of climate change impacts on the sediment cycle is currently a primary concern for environmental policy analysts in Mediterranean areas. Nevertheless, quantitative assessment of climate change impacts is still a complex task. The aim of this study was to implement a sediment model by taking advantage of sediment proxy information provided by reservoir bottom deposits and to use it for climate change assessment in a Mediterranean catchment. The sediment model was utilised in a catchment that drains into a large reservoir. The depositional history of the reservoir was reconstructed and used for sediment sub-model implementation. The model results were compared with gauged suspended sediment data in order to verify model robustness. Then, the model was coupled with future precipitation and temperature scenarios obtained from climate models. Climatological model outputs for two emission scenarios (A2 and B2) were simulated and the results compared with a reference scenario. Model results showed a general decrease in soil moisture and water discharge. Large floods, which are responsible for the majority of sediment mobilisation, also showed a general decrease. Sediment yield showed a clear reduction under the A2 scenario but increased under the B2 scenario. The computed specific sediment yield for the control period was 6.33 Mg ha−1 year−1, while for the A2 and B2 scenarios, it was 3.62 and 7.04 Mg ha−1 year−1, respectively. Furthermore, sediment transport showed an increase in its time compression, i.e. a stronger dependence of total sediment yield from the largest event contributions. This study shows a methodology for implementing a distributed sediment model by exploiting reservoir sedimentation volumes. This methodology can be applied to a wide range of catchments, given the high availability of reservoir sedimentation data. Moreover, this study showed how such a model can be used in the framework of a climate change study, providing a measure of the impact of climate change on soil erosion and sediment yields.

The promises and pitfalls of Stochastic Gradient Langevin Dynamics

Abstract: Stochastic Gradient Langevin Dynamics (SGLD) has emerged as a key MCMC algorithm for Bayesian learning from large scale datasets. While SGLD with decreasing step sizes converges weakly to the posterior distribution, the algorithm is often used with a constant step size in practice and has demonstrated spectacular successes in machine learning tasks. The current practice is to set the step size inversely proportional to N where N is the number of training samples. As N becomes large, we show that the SGLD algorithm has an invariant probability measure which significantly departs from the target posterior and behaves like as Stochastic Gradient Descent (SGD). This difference is inherently due to the high variance of the stochastic gradients. Several strategies have been suggested to reduce this effect; among them, SGLD Fixed Point (SGLDFP) uses carefully designed control variates to reduce the variance of the stochastic gradients. We show that SGLDFP gives approximate samples from the posterior distribution, with an accuracy comparable to the Langevin Monte Carlo (LMC) algorithm for a computational cost sublinear in the number of data points. We provide a detailed analysis of the Wasserstein distances between LMC, SGLD, SGLDFP and SGD and explicit expressions of the means and covariance matrices of their invariant distributions. Our findings are supported by limited numerical experiments.

Context-dependent kernel design for object matching and recognition

Abstract: The success of kernel methods including support vector networks (SVMs) strongly depends on the design of appropriate kernels. While initially kernels were designed in order to handle fixed-length data, their extension to unordered, variable-length data became more than necessary for real pattern recognition problems such as object recognition and bioinformatics. We focus in this paper on object recognition using a new type of kernel referred to as ldquocontext-dependentrdquo. Objects, seen as constellations of local features (interest points, regions, etc.), are matched by minimizing an energy function mixing (1) a fidelity term which measures the quality of feature matching, (2) a neighborhood criteria which captures the object geometry and (3) a regularization term. We will show that the fixed-point of this energy is a ldquocontext-dependentrdquo kernel (ldquoCDKrdquo) which also satisfies the Mercer condition. Experiments conducted on object recognition show that when plugging our kernel in SVMs, we clearly outperform SVMs with ldquocontext-freerdquo kernels.

The Impact of Human–Robot Interfaces on the Learning of Visual Objects

Abstract: This paper studies the impact of interfaces, allowing nonexpert users to efficiently and intuitively teach a robot to recognize new visual objects. We present challenges that need to be addressed for real-world deployment of robots capable of learning new visual objects in interaction with everyday users. We argue that in addition to robust machine learning and computer vision methods, well-designed interfaces are crucial for learning efficiency. In particular, we argue that interfaces can be key in helping nonexpert users to collect good learning examples and, thus, improve the performance of the overall learning system. Then, we present four alternative human–robot interfaces: Three are based on the use of a mediating artifact (smartphone, wiimote, wiimote and laser), and one is based on natural human gestures (with a Wizard-of-Oz recognition system). These interfaces mainly vary in the kind of feedback provided to the user, allowing him to understand more or less easily what the robot is perceiving and, thus, guide his way of providing training examples differently. We then evaluate the impact of these interfaces, in terms of learning efficiency, usability, and user’s experience, through a real world and large-scale user study. In this experiment, we asked participants to teach a robot 12 different new visual objects in the context of a robotic game. This game happens in a home-like environment and was designed to motivate and engage users in an interaction where using the system was meaningful. We then discuss results that show significant differences among interfaces. In particular, we show that interfaces such as the smartphone interface allows nonexpert users to intuitively provide much better training examples to the robot, which is almost as good as expert users who are trained for this task and are aware of the different visual perception and machine learning issues. We also show that artifact-mediated teaching is significantly more efficient for robot learning, and equally good in terms of usability and user’s experience, than teaching thanks to a gesture-based human-like interaction.