Convolutional neural networks are designed for dense data, but vision data is often sparse (stereo depth, point clouds, pen stroke, etc.). We present a method to handle sparse depth data with optional dense RGB, and accomplish depth completion and semantic segmentation changing only the last layer. Our proposal efficiently learns sparse features without the need of an additional validity mask. We show how to ensure network robustness to varying input sparsities. Our method even works with densities as low as 0.8% (8 layer lidar), and outperforms all published stateof-the-art on the Kitti depth completion benchmark.
We present research using the latest reinforcement learning algorithm for end-to-end driving without any mediated perception (object recognition, scene understanding). The newly proposed reward and learning strategies lead together to faster convergence and more robust driving using only RGB image from a forward facing camera. An Asynchronous Actor Critic (A3C) framework is used to learn the car control in a physically and graphically realistic rally game, with the agents evolving simultaneously on tracks with a variety of road structures (turns, hills), graphics (seasons, location) and physics (road adherence). A thorough evaluation is conducted and generalization is proven on unseen tracks and using legal speed limits. Open loop tests on real sequences of images show some domain adaption capability of our method.
Abstract-In this paper, we present our approach for collision risk estimation between vehicles. The vehicles are equipped with GPS receivers and communication devices. Our approach consists on using the knowledge given trough communication tool to predict the trajectories of the surrounding vehicles. Based on these trajectories, we identify the configurations of the collisions between vehicles. The risk is calculated using several indicators that are reflecting not only the possible collisions but also the dangerousness of these collisions. Our algorithm is tested on crossroads using scenarios involving real prototypes producing realistic scenarios.
International audienceVehicle localization (ground vehicles) is an important task for intelligent vehicle systems and vehicle cooperation may bring benefits for this task. A new cooperative multi-vehicle localization method using split covariance intersection filter is proposed in this paper. In the proposed method, each vehicle maintains an estimate of a decomposed group state and this estimate is shared with neighboring vehicles; the estimate of the decomposed group state is updated with both the sensor data of the ego-vehicle and the estimates sent from other vehicles; the covariance intersection filter which yields consistent estimates even facing unknown degree of inter-estimate correlation has been used for data fusion. A comparative study based simulations demonstrate the effectiveness and the advantage of the proposed cooperative localization method
Self-driving vehicles need to continuously analyse the driving scene, understand the behavior of other road users and predict their future trajectories in order to plan a safe motion and reduce their reaction time. Motivated by this idea, this paper addresses the problem of vehicle trajectory prediction over an extended horizon. On highways, human drivers continuously adapt their speed and paths according to the behavior of their neighboring vehicles. Therefore, vehicles' trajectories are very correlated and considering vehicle interactions makes motion prediction possible even before the start of a clear maneuver pattern. To this end, we introduce and analyze trajectory prediction methods based on how they model the vehicles interactions. Inspired by human reasoning, we use an attention mechanism that explicitly highlights the importance of neighboring vehicles with respect to their future states. We go beyond pairwise vehicle interactions and model higher order interactions. Moreover, the existence of different goals and driving behaviors induces multiple potential futures. We exploit a combination of global and partial attention paid to surrounding vehicles to generate different possible trajectory. Experiments on highway datasets show that the proposed model outperforms the state-of-the-art performances.
In this paper, we present a nonlinear adaptive controller for a two-vehicle automated overtaking maneuver. We consider the problem of an autonomous three-phase overtaking without the use of any roadway marking scheme or inter-vehicle communication. The developed feedback controller requires information for the current relative intervehicle position and orientation available from onboard sensors only. We apply standard robotic nomenclature for translational and rotational displacements and velocities and propose a general kinematic model of the vehicles during the overtaking maneuver including for the relative inter-vehicle kinematics. The overtaking maneuver is investigated as a tracking problem with respect to desired polynomial virtual trajectories for every phase, which are generated in real time. An update control law for the automated overtaking vehicle is designed that allows tracking the desired trajectories in the presence of unknown velocity of the overtaken vehicle. Simulation results illustrate the performance of the proposed controller.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.