ICRA 2021 论文列表

Recent studies have shown that enabling drones to change their morphology in flight can significantly increase their versatility in different tasks. In this paper, we investigate the aerodynamic effects caused by the partial overlap between the propellers and the main body of a morphing quadrotor during flight. We experimentally characterize such effects and design a morphology-aware control scheme to compensate them. We demonstrate the effectiveness of our approach by deploying the compensation scheme on a quadrotor that can fold its arms around the main body, comparing it against the same controller without the compensation scheme. Experimental results show that our compensation scheme can address the loss of thrust due to the overlap between the main body and the propellers, guaranteeing higher tracking accuracy, without requiring complex and computationally expensive aerodynamical models. To the best of our knowledge, this is the first work counteracting the aerodynamic effects of a morphing quadrotor during flight and showing the effects of partial overlap between a propeller and the central body of the drone.

Recently proposed DNN-based stereo matching methods that learn priors directly from data are known to suffer a drastic drop in accuracy in new environments. Although supervised approaches with ground truth disparity maps often work well, collecting them in each deployment environment is cumbersome and costly. For this reason, many unsupervised domain adaptation methods based on image-to-image translation have been proposed, but these methods do not preserve the geometric structure of a stereo image pair because the image-to-image translation is applied to each view separately. To address this problem, in this paper, we propose an attention mechanism that aggregates features in the left and right views, called Stereoscopic Cross Attention (SCA). Incorporating SCA to an image-to-image translation network makes it possible to preserve the geometric structure of a stereo image pair in the process of the image-to-image translation. We empirically demonstrate the effectiveness of the proposed unsupervised domain adaptation based on the image-to-image translation with SCA.

We provide a generic framework for the hand– eye calibration of vision-guided industrial robots. In contrast to traditional methods, we explicitly model the uncertainty of the robot in a statistically sound manner. Albeit the precision of modern industrial robots is high, their absolute accuracy typically is much lower. This uncertainty — if not considered — deteriorates the result of the hand–eye calibration. Our proposed framework not only results in a high accuracy of the computed hand–eye pose but also provides reliable information about the uncertainty of the robot. It further provides corrected robot poses for a convenient and inexpensive robot calibration. Our framework is generic in several respects: It supports the use of a calibration target as well as self-calibration without the need for known 3D points. It optionally allows the simultaneous calibration of the interior camera parameters. The framework is also generic with regard to the robot type, and hence supports articulated as well as SCARA robots, for example. Simulated and real experiments show the validity of the proposed methods.

We report the design of a morpho-functional robot called Husky Carbon. Our goal is to integrate two forms of mobility, aerial and quadrupedal legged locomotion, within a single platform. There are prohibitive design restrictions such as tight power budget and payload, which can particularly become important in aerial flights. To address these challenges, we pose a problem called the Mobility Value of Added Mass (MVAM) problem. In the MVAM problem, we attempt to allocate mass in our designs such that the energetic performance is affected the least. To solve the MVAM problem, we adopted a generative design approach using Grasshopper’s evolutionary solver to synthesize a parametric design space for Husky. Then, this space was searched for the morphologies that could yield a minimized Total Cost Of Transport (TCOT) and payload. This approach revealed that a front heavy quadrupedal robot can achieve a lower TCOT while retaining larger margins on allowable added mass to its design. Based on this framework Husky was built and tested as a front heavy robot.

For quadrotor trajectory planning, describing a polynomial trajectory through coefficients and end-derivatives both enjoy their own convenience in energy minimization. We name them double descriptions of polynomial trajectories. The transformation between them, causing most of the inefficiency and instability, is formally analyzed in this paper. Leveraging its analytic structure, we design a linear-complexity scheme for both jerk/snap minimization and parameter gradient evaluation, which possesses efficiency, stability, flexibility, and scalability. With the help of our scheme, generating an energy optimal (minimum snap) trajectory only costs 1 µs per piece at the scale up to 1,000,000 pieces. Moreover, generating large-scale energy-time optimal trajectories is also accelerated by an order of magnitude against conventional methods.

Manipulators can be added to legged robots, allowing them to interact with and change their environment. Legged mobile manipulation planners must consider how contact forces generated by these manipulators affect the system. Current planning strategies either treat these forces as immutable during planning or are unable to optimize over these contact forces while operating in real-time. This paper presents the Stability and Task Oriented Receding-Horizon Motion and Manipulation Autonomous Planner (STORMMAP) that is able to generate continuous plans for the robot’s motion and manipulation force trajectories that ensure dynamic feasibility and stability of the platform, and incentivizes accomplishing manipulation and motion tasks specified by a user. A variety of simulated experiments on a quadruped with a manipulator mounted to its torso demonstrate the versatility of STOR-MMAP. In contrast to existing state of the art methods, the approach described in this paper generates continuous plans in under ten milliseconds, an order of magnitude faster than previous strategies.

To perform manipulation tasks in the real world, robots need to operate on objects with various shapes, sizes and without access to geometric models. To achieve this it is often infeasible to train monolithic neural network policies across such large variations in object properties. Towards this generalization challenge, we propose to learn modular task policies which compose object-centric task-axes controllers. These task-axes controllers are parameterized by properties associated with underlying objects in the scene. We infer these controller parameters directly from visual input using multi-view dense correspondence learning. Our overall approach provides a simple and yet powerful framework for learning manipulation tasks. We empirically evaluate our approach on 3 different manipulation tasks and show its ability to generalize to large variance in object size, shape and geometry.

Point set registration (PSR) is an essential problem in communities of computer vision, medical robotics and biomedical engineering. This paper is motivated by considering the anisotropic characteristics of the error values in estimating both the positional and orientational vectors from the PSs to be registered. To do this, the multi-variate Gaussian and Kent distributions are utilized to model the positional and orientational uncertainties, respectively. Our contributions of this paper are three-folds: (i) the PSR problem using normal vectors is formulated as a maximum likelihood estimation (MLE) problem, where the anisotropic characteristics in both positional and normal vectors are considered; (ii) the matrix forms of the objective function and its associated gradients with respect to the desired parameters are provided, which can facilitate the computational process; (iii) two approaches of computing the normalizing constant in the Kent distribution are compared. We verify our proposed registration method on various PSs (representing pelvis and femur bones) in computer- assisted orthopedic surgery (CAOS). Extensive experimental results demonstrate that our method outperforms the state- of-the-art methods in terms of the registration accuracy and the robustness.

Robotics research has found numerous important applications of Riemannian geometry. Despite that, the concept remain challenging to many roboticists because the background material is complex and strikingly foreign. Beyond Riemannian geometry, there are many natural generalizations in the mathematical literature—areas such as Finsler geometry and spray geometry—but those generalizations are largely inaccessible, and as a result there remain few applications within robotics. This paper presents a re-derivation of spray and Finsler geometries, critical for the development of our recent work on geometric fabrics, which builds the ideas from familiar concepts in advanced calculus and the calculus of variations. We focus on the pragmatic and calculable results, avoiding the use of tensor notation to appeal to a broader audience and emphasizing geometric path consistency over ideas around connections and curvature. We hope that they will contribute to an increased understanding of generalized nonlinear, and even classical Riemannian, geometry within the robotics community and inspire future research into new applications.

Extensive efforts have been made to improve the generalization ability of Reinforcement Learning (RL) methods via domain randomization and data augmentation. However, as more factors of variation are introduced during training, optimization becomes increasingly challenging, and empirically may result in lower sample efficiency and unstable training. Instead of learning policies directly from augmented data, we propose SOft Data Augmentation (SODA), a method that decouples augmentation from policy learning. Specifically, SODA imposes a soft constraint on the encoder that aims to maximize the mutual information between latent representations of augmented and non-augmented data, while the RL optimization process uses strictly non-augmented data. Empirical evaluations are performed on diverse tasks from DeepMind Control suite as well as a robotic manipulation task, and we find SODA to significantly advance sample efficiency, generalization, and stability in training over state-of-the-art vision-based RL methods.1

GenGrid is a novel comprehensive open-source, distributed platform intended for conducting extensive swarm robotic experiments. The modular platform is designed to run swarm robotics experiments that are compatible with different types of mobile robots ranging from Colias, Kilobot, and E-puck [1]–[4]. The platform offers programmable control over the experimental setup and its parameters and acts as a tool to collect swarm robot data, including localization, sensory feedback, messaging, and interaction. GenGrid is designed as a modular grid of attachable computing nodes that offers bidirectional communication between the robotic agent and grid nodes and within grids. The paper describes the hardware and software architecture design of the GenGrid system. Further, it discusses some common experimental studies covering multi-robot and swarm robotics to showcase the platform’s use. GenGrid of 25 homogeneous cells with identical sensing and communication characteristics with a footprint of 37.5 cm × 37.5 cm, exhibits multiple capabilities with minimal resources. The open-source hardware platform is handy for running swarm experiments, including robot hopping based on multiple gradients, collective transport, shepherding, continuous pheromone deposition, and subsequent evaporation. The low-cost, modular, and open-source platform is significant in the swarm robotics research community, which is currently driven by commercial platforms that allow minimal modifications.

Vision-based tactile sensors have the potential to provide important contact geometry to localize the objective with visual occlusion. However, it is challenging to measure high-resolution 3D contact geometry for a compact robot finger, to simultaneously meet optical and mechanical constraints. In this work, we present the GelSight Wedge sensor, which is optimized to have a compact shape for robot fingers, while achieving high-resolution 3D reconstruction. We evaluate the 3D reconstruction under different lighting configurations, and extend the method from 3 lights to 1 or 2 lights. We demonstrate the flexibility of the design by shrinking the sensor to the size of a human finger for fine manipulation tasks. We also show the effectiveness and potential of the reconstructed 3D geometry for pose tracking in the 3D space.

Real-time semantic segmentation is a challenging task as both accuracy and inference speed need to be considered simultaneously. In real-world applications, it is usually achieved by deploying a deep neural network in modern GPU device. However, most of the work focused on real-time semantic segmentation is designed by significantly reducing computation complexity and model size. There are other factors that have a significant impact on inference speed are overlooked, especially when the network is running in modern GPU device. In this paper, we focus on designing a GPU-efficient network as backbone for real-time semantic segmentation. Dense connectivity can preserve and accumulate feature maps of multiple receptive fields and is therefore ideal for semantic segmentation. Therefore, we design a GPU-efficient network (DenseENet) with dense connectivity. The proposed DenseENet shows an obvious advantage in balancing accuracy and inference speed in modern GPU device. Specifically, on Cityscapes test set, DenseENet with a simple FCN decoder achieves 75.2% mIoU with 83.6 FPS for an input of 1024 × 2048 resolution and 73.6% mIoU with 132 FPS for an input of 768 × 1536 resolution on a single GTX 1080Ti card.

Object grasping in cluttered scenes is a widely investigated field of robot manipulation. Most of the current works focus on estimating grasp pose from point clouds based on an efficient single-shot grasp detection network. However, due to the lack of geometry awareness of the local grasping area, it may cause severe collisions and unstable grasp configurations. In this paper, we propose a two-stage grasp pose refinement network which detects grasps globally while fine-tuning low-quality grasps and filtering noisy grasps locally. Furthermore, we extend the 6-DoF grasp with an extra dimension as grasp width which is critical for collisionless grasping in cluttered scenes. It takes a single-view point cloud as input and predicts dense and precise grasp configurations. To enhance the generalization ability, we build a synthetic single-object grasp dataset including 150 commodities of various shapes, and a complex multi-object cluttered scene dataset including 100k point clouds with robust, dense grasp poses and mask annotations. Experiments conducted on Yumi IRB-1400 Robot demonstrate that the model trained on our dataset performs well in real environments and outperforms previous methods by a large margin.

Ground Penetrating Radar (GPR) is one of the most important non-destructive evaluation (NDE) instruments to detect and locate underground objects (i.e. rebars, utility pipes). Many of the previous researches focus on GPR image-based feature detection only, and none can process sparse GPR measurements to successfully reconstruct a very fine and detailed 3D model of underground objects for better visualization. To address this problem, this paper presents a novel robotic system to collect GPR data, localize the underground utilities, and reconstruct the underground objects’ dense point cloud model. This system is composed of three modules: 1) visual-inertial-based GPR data collection module which tags the GPR measurements with positioning information provided by an omnidirectional robot; 2) a deep neural network (DNN) migration module to interpret the raw GPR B-scan image into a cross-section of object model; 3) a DNN-based 3D reconstruction module, i.e., GPRNet, to generate underground utility model with the fine 3D point cloud. In this paper, both the quantitative and qualitative experiment results verify our method that can generate a dense and complete point cloud model of pipe-shaped utilities based on a sparse input, i.e., GPR raw data, with incompleteness and various noise. The experiment results on synthetic data as well as field test data further support the effectiveness of our approach.

Robotic audition is a basic sense that helps robots perceive the surroundings and interact with humans. Sound Source Localization (SSL) is an essential module for a robotic system. However, the performance of most sound source localization techniques degrades in noisy and reverberant environments due to inaccurate Time Difference of Arrival (TDoA) estimation. In robotic sound source localization, we are more interested in detecting the arrival of human speech than other sound sources. Ideally, we expect an effective TDoA estimation to respond only to speech signals, while masking off other interferences. In this paper, we propose a novel technique that learns to attend to speech fundamental frequency and harmonics while suppressing noise interference and reverberation. The novel TDoA feature is referred to as Generalized Cross Correlation with Phase Transform and Speech Mask (GCC-PHAT-SM). We perform sound source localization experiments on real-world data captured from a robotic platform. Experiments show that GCC-PHAT-SM feature significantly outperforms traditional Generalized Cross Correlation (GCC) feature in noisy and reverberant acoustic environments.

Testing mobile robots is difficult and expensive, and many faults go undetected. In this work we explore whether fuzzing, an automated test input generation technique, can more quickly find failure inducing inputs in mobile robots. We developed a simple fuzzing adaptation, BASE-FUZZ, and one specialized for fuzzing mobile robots, PHYS-FUZZ. PHYS-FUZZ is unique in that it accounts for physical attributes such as the robot dimensions, estimated trajectories, and time to impact measures to guide the test input generation process. The results of evaluating PHYS-FUZZ suggest that it has the potential to speed up the discovery of input scenarios that reveal failures, finding 56.5% more than uniform random input selection and 7.0% more than BASE-FUZZ during 7 days of testing.

The behavior of a robot is typically expressed as a set of source code files written using a programming language. As for any software engineering activity, programming robotic behaviors is a complex and error-prone task. This paper propose a methodology that aims to reduce the cost of producing a reliable software describing a robotic behavior by automatically testing it.We employ a fuzz testing technique to stress software components with randomly generated data. By applying fuzz testing to a complex robotic-software, we identified errors related to the coding, the way data is handled, the logic of the robotic behavior, and the initialization of architectural components. Furthermore, a panel of experts acquainted with the analyzed behavior have highlighted the relevance and the significance of our findings. Our fuzzer operates on the SMACH and ROS frameworks and it is available under the MIT public open source license.

The paper proposes a multi-modal sensor fusion algorithm that fuses WiFi, IMU, and floorplan information to infer an accurate and dense location history in indoor environments. The algorithm uses 1) an inertial navigation algorithm to estimate a relative motion trajectory from IMU sensor data; 2) a WiFi-based localization API in industry to obtain positional constraints and geo-localize the trajectory; and 3) a convolutional neural network to refine the location history to be consistent with the floorplan. We have developed a data acquisition app to build a new dataset with WiFi, IMU, and floorplan data with ground-truth positions at 4 university buildings and 3 shopping malls. Our qualitative and quantitative evaluations demonstrate that the proposed system is able to produce twice as accurate and a few orders of magnitude denser location history than the current standard, while requiring minimal additional energy consumption. We will publicly share our code and models.