ICRA 2018 论文列表

We propose a novel monocular visual odometry (VO) system called UnDeepVO in this paper. UnDeepVO is able to estimate the 6-DoF pose of a monocular camera and the depth of its view by using deep neural networks. There are two salient features of the proposed UnDeepVo:one is the unsupervised deep learning scheme, and the other is the absolute scale recovery. Specifically, we train UnDeepVoby using stereo image pairs to recover the scale but test it by using consecutive monocular images. Thus, UnDeepVO is a monocular system. The loss function defined for training the networks is based on spatial and temporal dense information. A system overview is shown in Fig. 1. The experiments on KITTI dataset show our UnDeepVO achieves good performance in terms of pose accuracy.

Coal mines pose a high safety risk for human workers. An autonomous inspection drone would enable a coal mine operation to reduce this risk by minimizing the time spent by workers inside the mine. This inspection drone must be highly robust to the harsh and dangerous environment of an underground coal mine, with high levels of coal dust and humidity that can obstruct many conventional sensing methods. For high functionality, the drone must sense and avoid potential obstacles and as well as inspect and map the mining wall face. The objective of this paper is to present ultra-wideband (UWB) radar as a robust sensing solution to this challenging environment and validate its performance experimentally in the typical coal mine environment, both statically and dynamically.

3D measurement of target objects characterized by specular reflection or subsurface scatterings cannot be measured by traditional 3D measurement methods because these targets have multiple light paths that make it difficult to determine the unique surface. We define these objects as complex light reflection objects. In this case, 3D measurement methods based on Light Transport (LT) Matrix estimation may be a solution to measure these complex light reflection objects, because LT Matrix captures every light path, and we can identify all 3D points on the target shape by using LT Matrix. However, these methods either provide low resolution results, or they are too slow for use in robot vision in practice. In this paper, we suppress the computational cost of LT Matrix estimation by dividing LT Matrix estimation into multi-scale. The proposed method reduces the number of candidate combinations between camera pixels and projector pixels greatly by using the information given by low resolution observations. The proposed algorithm allows high resolution measurement of the LT Matrix very efficiently. Furthermore, careful implementation of our method by using a sparse matrix representation achieves memory efficiency. We evaluated our method by measuring 3D points for a 256 × 256 resolution projector and camera system, which is an LT matrix 4096 times larger than that developed in our previous study [1] and 100 times faster than our naïve implementation of [2].

This paper introduces techniques for mosquito population surveys in the field using electrified screens (bug zappers) mounted to a UAV. Instrumentation on the UAV logs the UAV path and the GPS location, altitude, and time of each mosquito elimination. Hardware experiments with a UAV equipped with an electrified screen provide real-time measurements of (former) mosquito locations and mosquito-free volumes. Planning a trajectory for the UAV that maximizes the number of mosquito kills is related to the Traveling Salesman Problem, the Lawn Mower Problem and, most closely, Milling with Turn Cost. We reduce this problem to considering variants of covering a grid graph with minimum turn cost, corresponding to optimized energy consumption. We describe an exact method based on Integer Programming that is able to compute provably optimal instances with over 1,500 pixels. These solutions are then implemented on the UAV.

We present a new method to translate videos to commands for robotic manipulation using Deep Recurrent Neural Networks (RNN). Our framework first extracts deep features from the input video frames with a deep Convolutional Neural Networks (CNN). Two RNN layers with an encoder-decoder architecture are then used to encode the visual features and sequentially generate the output words as the command. We demonstrate that the translation accuracy can be improved by allowing a smooth transaction between two RNN layers and using the state-of-the-art feature extractor. The experimental results on our new challenging dataset show that our approach outperforms recent methods by a fair margin. Furthermore, we combine the proposed translation module with the vision and planning system to let a robot perform various manipulation tasks. Finally, we demonstrate the effectiveness of our framework on a full-size humanoid robot WALK-MAN.

Robots acting in open environments need to be able to handle novel objects. Based on the observation that objects within a category are often similar in their shapes and usage, we propose an approach for transferring grasping skills from known instances to novel instances of an object category. Correspondences between the instances are established by means of a non-rigid registration method that combines the Coherent Point Drift approach with subspace methods. The known object instances are modeled using a canonical shape and a transformation which deforms it to match the instance shape. The principle axes of variation of these deformations define a low-dimensional latent space. New instances can be generated through interpolation and extrapolation in this shape space. For inferring the shape parameters of an unknown instance, an energy function expressed in terms of the latent variables is minimized. Due to the class-level knowledge of the object, our method is able to complete novel shapes from partial views. Control poses for generating grasping motions are transferred efficiently to novel instances by the estimated non-rigid transformation.

In this work, we develop an approach for guiding robots to automatically localize and find the shapes of tumors and other stiff inclusions present in the anatomy. Our approach uses Gaussian processes to model the stiffness distribution and active learning to direct the palpation path of the robot. The palpation paths are chosen such that they maximize an acquisition function provided by an active learning algorithm. Our approach provides the flexibility to avoid obstacles in the robot's path, incorporate uncertainties in robot position and sensor measurements, include prior information about location of stiff inclusions while respecting the robot-kinematics. To the best of our knowledge this is the first work in literature that considers all the above conditions while localizing tumors. The proposed framework is evaluated via simulation and experimentation on three different robot platforms: 6-DoF industrial arm, da Vinci Research Kit (dVRK), and the Insertable Robotic Effector Platform (IREP). Results show that our approach can accurately estimate the locations and boundaries of the stiff inclusions while reducing exploration time.

We focus on a replanning scenario for quadrotors where considering time efficiency, non-static initial state and dynamical feasibility is of great significance. We propose a real-time B-spline based kinodynamic (RBK) search algorithm, which transforms a position-only shortest path search (such as A * and Dijkstra) into an efficient kinodynamic search, by exploring the properties of B-spline parameterization. The RBK search is greedy and produces a dynamically feasible time-parameterized trajectory efficiently, which facilitates non-static initial state of the quadrotor. To cope with the limitation of the greedy search and the discretization induced by a grid structure, we adopt an elastic optimization (EO) approach as a post-optimization process, to refine the control point placement provided by the RBK search. The EO approach finds the optimal control point placement inside an expanded elastic tube which represents the free space, by solving a Quadratically Constrained Quadratic Programming (QCQP) problem. We design a receding horizon replanner based on the local control property of B-spline. A systematic comparison of our method against two state-of-the-art methods is provided. We integrate our replanning system with a monocular vision-based quadrotor and validate our performance onboard.

In this paper we propose a novel general method to let a dynamical system fulfil at best a control task when the nominal parameters are not perfectly known. The approach is based on the introduction of the novel concept of closed-loop sensitivity, a quantity that relates parameter variations to deviations of the closed-loop trajectory of the system/controller pair. This new definition takes into account the dependency of the control inputs from the system states and nominal parameters as well as from the controller dynamics. The reference trajectory to be tracked is taken as optimization variable, and the dynamics of both the sensitivity and of its gradient are computed analytically along the system trajectories. We then show how this computation can be effectively exploited for solving trajectory optimization problems aimed at generating a reference trajectory that minimizes a norm of the closed-loop sensitivity. The theoretical results are validated via an extensive campaign of Monte Carlo simulations for two relevant robotic systems: a unicycle and a quadrotor UAV.

We present a deep neural network-based method to perform high-precision, robust and real-time 6 DOF positioning tasks by visual servoing. A convolutional neural network is fine-tuned to estimate the relative pose between the current and desired images and a pose-based visual servoing control law is considered to reach the desired pose. The paper describes how to efficiently and automatically create a dataset used to train the network. We show that this enables the robust handling of various perturbations (occlusions and lighting variations). We then propose the training of a scene-agnostic network by feeding in both the desired and current images into a deep network. The method is validated on a 6 DOF robot.

The most common paradigm for vision-based multi-object tracking is tracking-by-detection, due to the availability of reliable detectors for several important object categories such as cars and pedestrians. However, future mobile systems will need a capability to cope with rich human-made environments, in which obtaining detectors for every possible object category would be infeasible. In this paper, we address the problem of class-agnostic multi-object tracking using generic object proposals. We present an efficient segmentation mask-based tracker which associates pixel-precise masks reported by the segmentation. Our approach can utilize semantic information whenever it is available for classifying objects at the track level, while retaining the capability to track generic unknown objects in the absence of such information. We demonstrate experimentally that our approach achieves performance comparable to state-of-the-art tracking-by-detection methods for popular object categories such as cars and pedestrians. Additionally, we show that the proposed method can discover and robustly track a large variety of other objects.

Endovascular aneurysm repair (EVAR) is a minimally invasive approach for abdominal aortic aneurysm (AAA) treatment. Compared to open surgery, the benefits of EVAR include faster recovery and shorter time in hospital as well as no general anesthesia (in most cases). Though EVAR has become a preferred way to treat AAA with an increasing number of procedures, there are persisting complications, e.g. stent graft migration. Suturing the stent graft to the aorta increases the displacement force necessary to move the implant. This paper describes the design of a suturing catheter for EVAR. The suturing device consist of two modules which can be inserted through the femoral arteries into the abdominal aorta where both join using an electro-magnetic connector. The positioning module provides an anchor inside the aorta for the suturing module and new sequential positions for each stitch. Our large-scale prototype is validated inside a phantom vessel made of silicone material. We are able to successfully prove the concept of this novel single-sided suturing catheter for EVAR.

This paper presents the OTHex platform for aerial manipulation developed at LAAS-CNRS. The OTHex is probably the first multi-directional thrust platform designed to act as Flying Assistant which can aid human operators and/or Ground Manipulators to move long bars for assembly and maintenance tasks. The work emphasis is on task-driven custom design and experimental validations. The proposed control framework is built around a low-level geometric controller, and includes an external wrench estimator, an admittance filter, and a trajectory generator. This tool gives the system the necessary compliance to resist external force disturbances arising from contact with the surrounding environment or to parameter uncertainties in the load. A set of experiments validates the real-world applicability and robustness of the overall system.

In this paper, a control solution featuring an energetic constraint is developed to improve the safety of a robotic manipulator sharing its workspace with humans. This general control structure, exploits a generic safe controller that ensures the respect of multiple constraints thanks to a Linear Quadratic Problem formulation. With a unified energetic formulation, the controller allows to explicitly limit both the kinetic energy when moving and the wrench applied to the environment in case of contact with an unexpected obstacle. This control approach is experimented on a redundant Kuka LWR4+ robot which end-effector shall precisely point toward a given location while following a trajectory.

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 paper presents the first experimental results of crutch-less dynamic walking with paraplegics on a lower-body exoskeleton: ATALANTE, designed by the French start-up company Wandercraft. The methodology used to achieve these results is based on the partial hybrid zero dynamics (PHZD) framework for formally generating stable walking gaits. A direct collocation optimization formulation is used to provide fast and efficient generation of gaits tailored to each patient. These gaits are then implemented on the exoskeleton for three paraplegics. The end result is dynamically stable walking in an exoskeleton without the need for crutches. After a short period of tuning by the engineers and practice by the subjects, each subject was able to dynamically walk across a room of about 10 m up to a speed of 0.15 m/s (0.5 km/h) without the need for crutches or any other kind of assistance.

Developing a safe and efficient collision avoidance policy for multiple robots is challenging in the decentralized scenarios where each robot generates its paths without observing other robots' states and intents. While other distributed multi-robot collision avoidance systems exist, they often require extracting agent-level features to plan a local collision-free action, which can be computationally prohibitive and not robust. More importantly, in practice the performance of these methods are much lower than their centralized counterparts. We present a decentralized sensor-level collision avoidance policy for multi-robot systems, which directly maps raw sensor measurements to an agent's steering commands in terms of movement velocity. As a first step toward reducing the performance gap between decentralized and centralized methods, we present a multi-scenario multi-stage training framework to learn an optimal policy. The policy is trained over a large number of robots on rich, complex environments simultaneously using a policy gradient based reinforcement learning algorithm. We validate the learned sensor-level collision avoidance policy in a variety of simulated scenarios with thorough performance evaluations and show that the final learned policy is able to find time efficient, collision-free paths for a large-scale robot system. We also demonstrate that the learned policy can be well generalized to new scenarios that do not appear in the entire training period, including navigating a heterogeneous group of robots and a large-scale scenario with 100 robots. Videos are available at https://sites.google.com/view/drlmaca.

Motivated by the need for globally consistent tracking and mapping before autonomous robot navigation becomes realistically feasible, this paper presents a novel backend to monocular-inertial odometry. As some of the most challenging platforms for vision-based perception, we evaluate the performance of our system using Unmanned Aerial Vehicles (UAV s). Our experimental validation demonstrates that the proposed approach achieves drift correction and metric scale estimation from a single UAV on benchmarking datasets. Furthermore, the generality of our approach is demonstrated to achieve globally consistent maps built in a collaborative manner from two UAVs, each equipped with a monocular-inertial sensor suite, showing the possible gains opened by collaboration amongst robots to perform SLAM. Video - https://youtu.be/wbX36HBu2Eg.

Humans are adept at tool use. We can intuitively and immediately improvise and use unknown objects in our environment as tools, to assist us in performing tasks. In this study, we provide similar cognition and capabilities to robots. Neuroscientific studies on tool use have suggested that human dexterity with tools is enabled by the embodiment of the tools, which in effect, allows humans to immediately transfer prior skills acquired without tools, onto tasks requiring tool use. Here, utilizing the theoretical results from our investigations on embodiment and tool use in humans over the last years, we propose a concept and algorithm to enable similar skill transfer by robots. Our algorithm enables a robot that has had no prior learning with tools, to automatically recognize an object (seen for the first time) in its environment as a potential tool for an otherwise unattainable task, and use the tool to perform the task thereafter.