ICRA 2017 论文列表

Acquiring the accurate 3-D position of a target person around a robot provides fundamental and valuable information that is applicable to a wide range of robotic tasks, including home service, navigation and entertainment. This paper presents a real-time robotic 3-D human tracking system which combines a monocular camera with an ultrasonic sensor by the extended Kalman filter (EKF). The proposed system consists of three sub-modules: monocular camera sensor tracking model, ultrasonic sensor tracking model and multi-sensor fusion. An improved visual tracking algorithm is presented to provide partial location estimation (2-D). The algorithm is designed to overcome severe occlusions, scale variation, target missing and achieve robust re-detection. The scale accuracy is further enhanced by the estimated 3-D information. An ultrasonic sensor array is employed to provide the range information from the target person to the robot and Gaussian Process Regression is used for partial location estimation (2-D). EKF is adopted to sequentially process multiple, heterogeneous measurements arriving in an asynchronous order from the vision sensor and the ultrasonic sensor separately. In the experiments, the proposed tracking system is tested in both simulation platform and actual mobile robot for various indoor and outdoor scenes. The experimental results show the superior performance of the 3-D tracking system in terms of both the accuracy and robustness.

In arc welding processes, real time control algorithms have to be developed in order to achieve desired weld quality. However, there could exist big uncertainties and noise in the process, which nullifies the conventional online control method. Besides, due to the modelling difficulty and low experimental efficiency, this task is usually performed offline. In this paper, a real time parameter optimization method is developed to find the optimal welding parameters to achieve desired characteristic performance. Gaussian Process Regression (GPR), a non-parametric modelling technique, is employed to model the relationship between input welding parameters and output characteristic performance. The GPR surrogated Bayesian Optimization Algorithm (GPRBOA) is proposed to optimize the welding parameters. Lower Confidence Bound (LCB) and Upper Confidence Bound (UCB) acquisition functions are utilized. Gas tungsten arc welding experiments were performed and the corresponding experimental data are collected and utilized to validate the proposed modelling method. The predicted characteristic performance is compared with the original data and it shows that the modelling method can accurately predict the weld bead geometry. The control algorithms were demonstrated and the results are presented. This paper opens a door for real time parameter tuning to achieve desired performance, hence the proposed method will innovate the arc welding process.

Changes to robotic systems as they are updated or upgraded often affect the flow of control and sensor data. Developers and users spend a significant amount of time tracing the impact of these changes that could otherwise have negative impacts on the robot's performance and behavior. Changes to the rates at which data is published from sensors, controllers, and other parts of the system are particularly subtle and difficult to detect. These rate changes, even if minor (e.g. lowering the frame rate of a camera), can propagate throughout the system and have broad impacts. In this work, we develop and implement an approach to help identify the set of components whose rate may be impacted by a system change. The approach builds on the insight that certain code patterns render component's outgoing data rate independent of the component's incoming data rate. We use that insight to reduce the number of components reported as affected by the change to minimize the number of components that must be reevaluated by the developer. A study of an implementation of the approach on three ROS systems shows that it can reduce the size of the impact set by up to 41% in cases when the changes have broad data impacts. The analysis is performed at compile time and only adds a third more to the compilation time.

Inverse Reinforcement Learning (IRL) for path planning enables robots to learn cost functions for difficult tasks from demonstration, instead of hard-coding them. However, IRL methods face practical limitations that stem from the need to repeat costly planning procedures. In this paper, we propose Rapidly Exploring Learning Trees (RLT*), which learns the cost functions of Optimal Rapidly Exploring Random Trees (RRT*) from demonstration, thereby making inverse learning methods applicable to more complex tasks. Our approach extends Maximum Margin Planning to work with RRT* cost functions. Furthermore, we propose a caching scheme that greatly reduces the computational cost of this approach. Experimental results on simulated and real-robot data from a social navigation scenario show that RLT* achieves better performance at lower computational cost than existing methods. We also successfully deploy control policies learned with RLT* on a real telepresence robot.

Maximizing information gathered within a budget is a relevant problem for information gathering tasks for robots with cost or operating time constraints. This problem is also known as the informative path planning (IPP) problem or correlated orienteering. It can be formalized as that of finding budgeted routes in a graph such that the reward collected by the route is maximized, where the reward at nodes can be dependent. Unfortunately, the problem is NP-Hard and the state of the art methods are too slow to even present an approximate solution online. Here we present Randomized Anytime Orienteering (RAOr) algorithm that provides near optimal solutions while demonstrably converging to an efficient solution in runtimes that allows the solver to be run online. The key idea of our approach is to pose orienteering as a combination of a Constraint Satisfaction Problem and a Traveling Salesman Problem. This formulation allows us to restrict the search space to routes that incur minimum distance to visit a set of selected nodes, and rapidly search this space using random sampling. The paper provides the analysis of asymptotic near-optimality, convergence rates for RAOr algorithms, and present strategies to improve anytime performance of the algorithm. Our experimental results suggest an improvement by an order of magnitude over the state of the art methods in relevant simulation and in real world scenarios.

This work addresses the task of camera localization in a known 3D scene given a single input RGB image. State-of-the-art approaches accomplish this in two steps: firstly, regressing for every pixel in the image its 3D scene coordinate and subsequently, using these coordinates to estimate the final 6D camera pose via RANSAC. To solve the first step. Random Forests (RFs) are typically used. On the other hand. Neural Networks (NNs) reign in many dense regression tasks, but are not test-time efficient. We ask the question: which of the two is best for camera localization? To address this, we make two method contributions: (1) a test-time efficient NN architecture which we term a ForestNet that is derived and initialized from a RF, and (2) a new fully-differentiable robust averaging technique for regression ensembles which can be trained end-to-end with a NN. Our experimental findings show that for scene coordinate regression, traditional NN architectures are superior to test-time efficient RFs and ForestNets, however, this does not translate to final 6D camera pose accuracy where RFs and ForestNets perform slightly better. To summarize, our best method, a ForestNet with a robust average, which has an equivalent fast and lightweight RF, improves over the state-of-the-art for camera localization on the 7-Scenes dataset [1]. While this work focuses on scene coordinate regression for camera localization, our innovations may also be applied to other continuous regression tasks.

In this paper, we present a monocular direct semi-dense SLAM (Simultaneous Localization And Mapping) method that can handle both radial distortion and rolling-shutter distortion. Such distortions are common in, but not restricted to, situations when an inexpensive wide-angle lens and a CMOS sensor are used, and leads to significant inaccuracy in the map and trajectory estimates if not modeled correctly. The apparent naive solution of simply undistorting the images using pre-calibrated parameters does not apply to this case since rows in the undistorted image are no longer captured at the same time. To address this we develop an algorithm that incorporates radial distortion into an existing state-of-the-art direct semi-dense SLAM system that takes rolling-shutters into account. We propose a method for finding the generalized epipolar curve for each rolling-shutter radially distorted image. Our experiments demonstrate the efficacy of our approach and compare it favorably with the state-of-the-art in direct semi-dense rolling-shutter SLAM.

We present a new image feature detection method. Our method selects features based on segmenting points with high local intensity variations across different scales using a robust rank order statistics approach. Our method produces a large number of repeatable features that are invariant to several image transformations such as rotation, scaling, viewpoint, and lighting variations. We show the advantages of our feature in comparison to other existing features using the Oxford dataset. We also show that, when used in monocular and stereo SLAM systems, our feature outperforms SIFT in terms of the pose estimation accuracy using several public datasets including the KITTI dataset.

This paper presents a novel appearance and shape feature, RISAS, which is robust to viewpoint, illumination, scale and rotation variations. RISAS consists of a keypoint detector and a feature descriptor both of which utilise texture and geometric information present in the appearance and shape channels. A novel response function based on the surface normals is used in combination with the Harris corner detector for selecting keypoints in the scene. A strategy that uses the depth information for scale estimation and background elimination is proposed to select the neighbourhood around the keypoints in order to build precise invariant descriptors. Proposed descriptor relies on the ordering of both grayscale intensity and shape information in the neighbourhood. Comprehensive experiments which confirm the effectiveness of the proposed RGB-D feature when compared with CSHOT [1] and LOIND[2] are presented. Furthermore, we highlight the utility of incorporating texture and shape information in the design of both the detector and the descriptor by demonstrating the enhanced performance of CSHOT and LOIND when combined with RISAS detector.

Visual SLAM in low illumination scenes remains a considerably challenging task since the available amount of appearance information frequently stays insufficient. To tackle with this problem, we propose a novel SLAM framework by using both appearance information and thermal information, which possesses illumination-free recognizable contents, in a flexible manner. The key idea is to continuously update a RGB-T map, which contains both RGB and thermal map points to implement location and mapping. More specifically, in our SLAM system, we detect features in both RGB and thermal images and combine them together to update the RGB-T map and implement simultaneous location and mapping. Both quantitative and qualitative results demonstrate the effectiveness of our framework, especially under low illumination environments.

The SLAM problem is known to have a special property that when robot orientation is known, estimating the history of robot poses and feature locations can be posed as a standard linear least squares problem. In this work, we develop a SLAM framework that uses relative feature-to-feature measurements to exploit this structural property of SLAM. Relative feature measurements are used to pose a linear estimation problem for pose-to-pose orientation constraints. This is followed by solving an iterative non-linear on-manifold optimization problem to compute the maximum likelihood estimate for robot orientation given relative rotation constraints. Once the robot orientation is computed, we solve a linear problem for robot position and map estimation. Our approach reduces the computational complexity of non-linear optimization by posing a smaller optimization problem as compared to standard graph-based methods for feature-based SLAM. Further, empirical results show our method avoids catastrophic failures that arise in existing methods due to using odometery as an initial guess for non-linear optimization, while its accuracy degrades gracefully as sensor noise is increased. We demonstrate our method through extensive simulations and comparisons with an existing state-of-the-art solver.

The integration of minimally invasive robotic assistance and image-guidance can have positive impact on joint fracture surgery, providing a better clinical outcome with respect to the current open procedure. In this paper, a new design of the RAFS surgical system is presented. The redesign of the robotic system and its integration with a novel 3D navigation system through a new clinical workflow, overcomes the drawbacks of the earlier prototype. This makes the RAFS surgical system more suitable to clinical scenarios in the operating theatre. System accuracy and effectiveness are successfully demonstrated through laboratory trials and preliminary cadaveric trials. The experimental results demonstrate that the system allows the surgeon to reduce a 2-fragment distal femur fracture in a cadaveric specimen, with a reduction accuracy of up to 0.85 mm and 2.2°. Preliminary cadaveric trials also provided a positive and favorable outcome pointing to the usability and safety of the RAFS system in the operating theatre, potentially enhancing the capacity of joint fracture surgeries.

Robot calibration and performance will degrade if proper maintenance isn't performed. There have been challenges for manufacturers to optimize the maintenance strategy and minimize unexpected shutdowns. Prognostics and health management (PHM) can be applied to industrial robots through the development of performance metrics, test methods, reference datasets, and supporting tools. A subset of this research involves developing a quick health assessment methodology emphasizing the identification of the positional health (position and orientation accuracy) changes. This methodology enables manufacturers to quickly assess the static/dynamic position and orientation accuracies of their robot systems. In this paper, the National Institute of Standards and Technology's (NIST) effort to develop the measurement science to support this development is presented, including the modeling and algorithm development for the test method, the advanced sensor development to measure 7-D information (time, X, Y, Z, roll, pitch, and yaw), algorithms to analyze the data, and a use case to present the results.

This paper contributes to quantifying the notion of robotic fitness by developing a set of necessary conditions that determine whether a small quadruped has the ability to open a class of doors or climb a class of stairs using only quasi-static maneuvers. After verifying that several such machines from the recent robotics literature are mismatched in this sense to the common human scale environment, we present empirical work-arounds for the Minitaur quadrupedal platform that enable it to leap up, force the door handle and push through the door, as well as bound up the stairs, thereby accomplishing through dynamical maneuvers otherwise (i.e., quasi-statically) unachievable tasks.

Quadrupedal locomotion can be described as a constrained optimization problem that is very hard to solve due to the high dimensional, nonlinear and non-smooth system dynamics. In this paper, we propose a formulation that can be solved within few seconds using sequential quadratic programming. This method considers only a simplified model that just sufficiently represents the system dynamics. The output is a very coarse plan, which can be accurately and robustly followed on a real system using hierarchical whole-body control combined with inverted pendulum-based reactive stepping. Using the fully torque controllable quadrupedal robot ANYmal, we present successful experiments for walking, trotting, and gait transitions even under substantial external disturbances.

In this paper, we present an optimization-based framework for generating quadrotor trajectories which are free of collision in dynamic environments with both static and moving obstacles. Using the finite-horizon motion prediction of moving obstacles, our method is able to generate safe and smooth trajectories with minimum control efforts. Our method optimizes trajectories globally for all observed moving and static obstacles, such that the avoidance behavior is most unnoticeable. This method first utilizes semi-definite relaxation on a quadratically constrained quadratic programming (QCQP) problem to eliminate the nonconvex constraints in the moving obstacle avoidance problem. A feasible and reasonably good solution to the original nonconvex problem is obtained using a randomization method and convex linear restriction. We detail the trajectory generation formulation and the solving procedure of the nonconvex quadratic program. Our approach is validated by both simulation and experimental results.

Collisions between quadrotor UAVs and the environment often occur, for instance, under faulty piloting, from wind gusts, or when obstacle avoidance fails. Airspace regulations are forcing drone companies to build safer drones; many quadrotor drones now incorporate propeller protection. However, propeller protected quadrotors still do not detect or react to collisions with objects such as walls, poles and cables. In this paper, we present a collision recovery pipeline which controls propeller protected quadrotors to recover from collisions. This pipeline combines concepts from impact dynamics, fuzzy logic, and aggressive quadrotor attitude control. The strategy is validated via a comprehensive Monte Carlo simulation of collisions against a wall, showing the feasibility of recovery from challenging collision scenarios. The pipeline is implemented on a custom experimental quadrotor platform, demonstrating feasibility of real-time performance and successful recovery from a range of pre-collision conditions. The ultimate goal of the research is to implement a general collision recovery solution as a safety feature for quadrotor flight controllers.

We address the problem of humanoid falling with a decoupled strategy consisting of a pre-impact and a postimpact stage. In the pre-impact stage, geometrical reasoning allows the robot to choose appropriate impact points in the surrounding environment and to adopt a posture to reach them while avoiding impact-singularities and preparing for the postimpact. The surrounding environment can be unstructured and may contain cluttered obstacles. The post-impact stage uses a quadratic program controller that adapts on-line the joint proportional-derivative (PD) gains to make the robot compliant-to absorb impact and post-impact dynamics, which lowers possible damage risks. This is done by a new approach incorporating the stiffness and damping gains directly as decision variables in the QP along with the usually-considered variables of joint accelerations and contact forces. Constraints of the QP prevent the motors from reaching their torque limits during the fall. Several experiments on the humanoid robot HRP-4 in a full-dynamics simulator are presented and discussed.

The Portable Robotic Optical/Tactile ObservatioN PACKage (PROTONPACK, or Proton for short) is a new handheld visuo-haptic sensing system that records surface interactions. We previously demonstrated system calibration and a classification task using external motion tracking. This paper details improvements in surface classification performance and removal of the dependence on external motion tracking, necessary before embarking on our goal of gathering a vast surface interaction dataset. Two experiments were performed to refine data collection parameters. After adjusting the placement and filtering of the Proton's high-bandwidth accelerometers, we recorded interactions between two differently-sized steel tooling ball end-effectors (diameter 6.35 and 9.525 mm) and five surfaces. Using features based on normal force, tangential force, end-effector speed, and contact vibration, we trained multi-class SVMs to classify the surfaces using 50 ms chunks of data from each end-effector. Classification accuracies of 84.5% and 91.5% respectively were achieved on unseen test data, an improvement over prior results. In parallel, we pursued on-board motion tracking, using the Proton's camera and fiducial markers. Motion tracks from the external and onboard trackers agree within 2 mm and 0.01 rad RMS, and the accuracy decreases only slightly to 87.7% when using onboard tracking for the 9.525 mm end-effector. These experiments indicate that the Proton 2 is ready for portable data collection.