IROS 2021 论文列表

Humans learn complex motor skills with practice and training. Though the learning process is not fully understood, several theories from motor learning, neuroscience, education, and game design suggest that curriculum-based training may be the key to efficient skill acquisition. However, designing such a curriculum and understanding its effects on learning are challenging problems. In this paper, we define the Human-skill Curriculum Markov Decision Process (H-CMDP) to systematize the design of training protocols. We also identify a vocabulary of performance features to enable the approximation for a human’s skill level across a variety of cognitive and motor tasks. A novel task domain is introduced as a testbed to evaluate the effectiveness of our approach. Human subject experiments show that (1) participants can learn to improve their performance in tasks within this domain, (2) the learning is quantifiable via our performance features, and (3) the domain is flexible enough to create distinct levels of difficulty. The long-term goal of this work is to systematize the process of curriculum-based training toward the design of protocols for robot-mediated rehabilitation.

We investigate the capacitated vehicle routing problem (CVRP) under a robotics context, where a vehicle with limited payload must complete delivery (or pickup) tasks to serve a set of geographically distributed customers with varying demands. In classical CVRP, a customer location is modeled as a point. In many robotics applications, however, it is more appropriate to model such "customer locations" as 2D regions. For example, in aerial delivery, a drone may drop a package anywhere in a customer’s lot. This yields the problem of CVRG (Capacitated Vehicle Routing with Target Geometric Constraints). Computationally, CVRP is already strongly NPhard; CVRG is therefore more challenging. Nevertheless, we develop fast algorithms for CVRG, capable of computing high quality solutions for hundreds of regions. Our algorithmic solution is guaranteed to be optimal when customer regions are convex. Numerical evaluations show that our proposed methods significantly outperform greedy best-first approaches. Comprehensive simulation studies confirm the effectiveness of our methods.

Traditionally vibrotactile feedback delivered by haptic interfaces is used to provide additional support for visual interaction via prehensile object manipulation using fingers. Nevertheless, haptic stimuli can be also applied for non-prehensile interaction that involves movements of the elbow joint. In this paper, we have designed a table-top haptic device that provides vibrotactile stimulation to the user’s fingertips. This stimulation creates a kinesthetic illusion of the user’s forearm pivot movement with respect to the elbow joint. The vibrotactile stimulation is amplified by the pseudo-haptic effect delivered through a head-mounted display (HMD). The efficacy of the approach was validated in experiments with human subjects. The results show that the combination of pseudo-haptic and haptic illusion effects can be used to render various soft and rigid virtual objects.

Learning to distinguish independent moving objects from the observed optical flow with a moving camera remains challenging. In this work, we first present a novel camera pose compensation (CPC) scheme. With the help of ingenious geometric analysis, it breaks the observed optical flow into patterns that are easier to interpret for the motion segmentation network. Secondly, we further refine such compensation with a camera parameter aware (CPA) module to account for poses’ errors in the CPC processing and enhance the entire network’s tolerance to noises. Additionally, an MMPNet is developed to intensify the identification ability of overall motion patterns. It reaches a larger receptive field with a bottom-up information transmission structure and integrates motion information at different granularities. We demonstrate the benefits of our framework on FlyingThings3D and Monkaa datasets. Without the complement of semantic information, our approach outperforms the top methods for moving objects segmentation.

Imitation learning (IL) algorithms have shown promising results for robots to learn skills from expert demonstrations. However, they need multi-task demonstrations to be provided at once for acquiring diverse skills, which is difficult in real world. In this work we study how to realize continual imitation learning ability that empowers robots to continually learn new tasks one by one, thus reducing the burden of multitask IL and accelerating the process of new task learning at the same time. We propose a novel trajectory generation model that employs both a generative adversarial network and a dynamics-aware prediction model to generate pseudo trajectories from all learned tasks in the new task learning process. Our experiments on both simulation and real-world manipulation tasks demonstrate the effectiveness of our method.

High quality object proposals are crucial in visual tracking algorithms that utilize region proposal network (RPN). Refinement of these proposals, typically by box regression and classification in parallel, has been popularly adopted to boost tracking performance. However, it still meets problems when dealing with complex and dynamic background. Thus motivated, in this paper we introduce an improved proposal refinement module, Cascaded Regression-Align-Classification (CRAC), which yields new state-of-the-art performances on many benchmarks.First, having observed that the offsets from box regression can serve as guidance for proposal feature refinement, we design CRAC as a cascade of box regression, feature alignment and box classification. The key is to bridge box regression and classification via an alignment step, which leads to more accurate features for proposal classification with improved robustness. To address the variation in object appearance, we introduce an identification-discrimination component for box classification, which leverages offline reliable fine-grained template and online rich background information to distinguish the target from background. Moreover, we present pyramid RoIAlign that benefits CRAC by exploiting both the local and global cues of proposals. During inference, tracking proceeds by ranking all refined proposals and selecting the best one. In experiments on seven benchmarks including OTB-2015, UAV123, NfS, VOT-2018, TrackingNet, GOT-10k and LaSOT, our CRACT exhibits very promising results in comparison with state-of-the-art competitors and runs in real-time at 28 fps.

Pedestrian trajectory prediction is a challenge because of the complex social interactions in context and the elusive intention of each pedestrian. Collision avoidance is one of the most common social interactions in real world, while existing data-driven works have not handled it well yet. In order to address this issue, we propose a framework that considers the theory about the minimum distance between each pedestrian-pedestrian pair and the corresponding time as the collision related prior knowledge. With the prior, our social refinement module, called Collision-prior Guided Refinement, can be guided to understand the collision situations of a crowd through a message passing mechanism. To focus on more useful information from context, we also introduce pedestrian-wise attention and collision gate to jointly judge collision potential for all pedestrian-pedestrian pairs. Experimental results demonstrate that our framework can achieve competitive results on ETH and UCY datasets by comparing with existing works. In addition, it indicates the superiority of our framework in the aspect of collision avoidance.

Road curb detection is important for autonomous driving. It can be used to determine road boundaries to constrain vehicles on roads, so that potential accidents could be avoided. Most of the current methods detect road curbs online using vehicle-mounted sensors, such as cameras or 3-D Lidars. However, these methods usually suffer from severe occlusion issues. Especially in highly-dynamic traffic environments, most of the field of view is occupied by dynamic objects. To alleviate this issue, we detect road curbs offline using high-resolution aerial images in this paper. Moreover, the detected road curbs can be used to create high-definition (HD) maps for autonomous vehicles. Specifically, we first predict the pixel-wise segmentation map of road curbs, and then conduct a series of post-processing steps to extract the graph structure of road curbs. To tackle the disconnectivity issue in the segmentation maps, we propose an innovative connectivity-preserving loss (CP-loss) to improve the segmentation performance. The experimental results on a public dataset demonstrate the effectiveness of our proposed loss function. This paper is accompanied with a demonstration video and a supplementary document, which are available at https://sites.google.com/view/cp-loss.

This paper considers online object-level mapping using partial point-cloud observations obtained online in an unknown environment. We develop an approach for fully Convolutional Object Retrieval and Symmetry-AIded Registration (CORSAIR). Our model extends the Fully Convolutional Geo-metric Features model to learn a global object-shape embedding in addition to local point-wise features from the point-cloud observations. The global feature is used to retrieve a similar object from a category database, and the local features are used for robust pose registration between the observed and the retrieved object. Our formulation also leverages symmetries, present in the object shapes, to obtain promising local-feature pairs from different symmetry classes for matching. We present results from synthetic and real-world datasets with different object categories to verify the robustness of our method.

Place recognition is indispensable for a drift-free localization system. Due to the variations of the environment, place recognition using single-modality has limitations. In this paper, we propose a bi-modal place recognition method, which can extract a compound global descriptor from the two modalities, vision and LiDAR. Specifically, we first build the elevation image generated from 3D points as a structural representation. Then, we derive the correspondences between 3D points and image pixels that are further used in merging the pixel-wise visual features into the elevation map grids. In this way, we fuse the structural features and visual features in the consistent bird-eye view frame, yielding a semantic representation, namely CORAL. And the whole network is called CORAL-VLAD. Comparisons on the Oxford RobotCar show that CORAL-VLAD has superior performance against other state-of-the-art methods. We also demonstrate that our network can be generalized to other scenes and sensor configurations on cross-city datasets.

Semantic segmentation serves as a cornerstone for safety autonomous driving and has been achieved remarkable progress at the price of dense annotations. Unsupervised domain adaptation was widely utilized to addresses this labor-intensive problem, which transfers the knowledge learned from labeled synthetic datset to real-world without any annotations. However, most existing adaptation works predict the segmentation results and domain identification results separately only with the last-layer feature, and ignore the intrinsic relationship among these two tasks. To address this issue, we present a CO-Interactive Network (COINet) for unsupervised adaptive segmentation. In particular, we propose a scale-aware distilled decoder to integrate multi-scale features dynamically through the designed inter-distilled module (IDM) and obtain fine-grained feature representations. A dual-task classifier is advanced with this decoder, to jointly predict the segmentation results and pixel-wise domain prediction results, which extracts shared complementary information for accurate segmentation. We further devise a co-interactive loss to explicitly model the intrinsic relationship among the segmentation and domain prediction, enabling the feature distribution alignment in pixel-level and an optimal segmentation decision boundary. We demonstrate the effectiveness of the proposed COINet on benchmark adaptation settings with extensive experimental and ablation results, and our model shows favorable performance against existing algorithms.

General contact-rich manipulation problems are long-standing challenges in robotics due to the difficulty of understanding complicated contact physics. Deep reinforcement learning (RL) has shown great potential in solving robot manipulation tasks. However, existing RL policies have limited adaptability to environments with diverse dynamics properties, which is pivotal in solving many contact-rich manipulation tasks. In this work, we propose Contact-aware Online COntext Inference (COCOI), a deep RL method that encodes a context embedding of dynamics properties online using contact-rich interactions. We sample sensor data using a novel contact-aware strategy and formulate an interpretable dynamics transition module. We study this method based on a novel and challenging non-planar pushing task, where the robot uses a monocular camera image and wrist force torque sensor reading to push an object to a goal location while keeping it upright. We run extensive experiments to demonstrate the capability of COCOI in a wide range of settings and dynamics properties in simulation, and also in a sim-to-real transfer scenario on a real robot (Webpage: https://context-inference.github.io/).

LiDAR map matching(LMM) is a critical localization technique in autonomous driving while existing methods have problems in terms of both accuracy and robustness when driving in the scenes with poor structure information (e.g. highways). This paper put forward a multi-level intensity map based cascaded network for LiDAR map matching in autonomous driving. The network uses an effective multi-level intensity map representation to compactly encode the appearance and structure information of point clouds, which effectively reduce the position ambiguity in structure-less scenarios. Besides, this method leverages the multi-scale nature of deep neural networks and matches the online LiDAR observation with the offline map in a coarse-to-fine manner so as to balance the time-consuming and precision. Extensive experiments on diverse autonomous driving environments demonstrate the superiority of our proposed method over other existing state-of-the-art methods.

In this paper, we propose a highly accurate continuous-time trajectory estimation framework dedicated to SLAM (Simultaneous Localization and Mapping) applications, which enables fuse high-frequency and asynchronous sensor data effectively. We apply the proposed framework in a 3D LiDAR-inertial system for evaluations. The proposed method adopts a non-rigid registration method for continuous-time trajectory estimation and simultaneously removing the motion distortion in LiDAR scans. Additionally, we propose a two-state continuous-time trajectory correction method to efficiently and efficiently tackle the computationally-intractable global optimization problem when loop closure happens. We examine the accuracy of the proposed approach on several publicly available datasets and the data we collected. The experimental results indicate that the proposed method outperforms the discrete-time methods regarding accuracy especially when aggressive motion occurs. Furthermore, we open source our code at https://github.com/APRIL-ZJU/clins to benefit research community.

We propose a vision-based reinforcement learning (RL) approach for closed-loop trajectory generation in an arm reaching problem. Arm trajectory generation is a fundamental robotics problem which entails finding collision-free paths to move the robot’s body (e.g. arm) in order to satisfy a goal (e.g. place end-effector at a point). While classical methods typically require the model of the environment to solve a planning, search or optimization problem, learning-based approaches hold the promise of directly mapping from observations to robot actions. However, learning a collision-avoidance policy using RL remains a challenge for various reasons, including, but not limited to, partial observability, poor exploration, low sample efficiency, and learning instabilities. To address these challenges, we present a residual-RL method that leverages a greedy goal-reaching RL policy as the base to improve exploration, and the base policy is augmented with residual state-action values and residual actions learned from images to avoid obstacles. Further more, we introduce novel learning objectives and techniques to improve 3D understanding from multiple image views and sample efficiency of our algorithm. Compared to RL baselines, our method achieves superior performance in terms of success rate.

This paper presents an accurate, highly efficient and learning free method for large-scale radar odometry estimation. By using a simple filtering technique that keeps the strongest returns, we produce a clean radar data representation and reconstruct surface normals for efficient and accurate scan matching. Registration is carried out by minimizing a point-to-line metric and robustness to outliers is achieved using a Huber loss. Drift is additionally reduced by jointly registering the latest scan to a history of keyframes. We found that our odometry pipeline generalize well to different sensor models and datasets without changing a single parameter. We evaluate our method in three widely different environments and demonstrate an improvement over spatially cross validated state-of-the-art with an overall translation error of 1.76% in a public urban radar odometry benchmark, running merely on a single laptop CPU thread at 55 Hz.

A Flying-type cable climbing robot, CCRobot-IV-F, is presented in this paper. It is a climbing precursor of the fourth version of CCRobot, designed to surpass the abilities of previous robots with high climbing speed and obstacle-crossing capability. CCRobot-IV-F weighs less than 10 kg and a no-load speed of up to 4.5 m/s, which significantly exceeds that of other climbing robots. A dynamic model integrated with a cable-fixed coordinate system is developed, and a cascaded controller designed for stabilizing hover and climb with grippers, when a Global Positioning System and magnetometer are unavailable, is shown to work reliably in practice. Experimental results show that CCRobot-IV-F significantly improves the locomotive performance of CCRobot-IV, exhibiting fast speed, good payload capacity, and excellent obstacle-crossing capability. Moreover, CCRobot-IV-F is applied to a cable-stayed bridge in the field.

Tracking the 6D pose of objects in video sequences is important for robot manipulation. Most prior efforts, however, often assume that the target object's CAD model, at least at a category-level, is available for offline training or during online template matching. This work proposes BundleTrack, a general framework for 6D pose tracking of novel objects, which does not depend upon 3D models, either at the instance or category-level. It leverages the complementary attributes of recent advances in deep learning for segmentation and robust feature extraction, as well as memory-augmented pose graph optimization for spatiotemporal consistency. This enables long-term, low-drift tracking under various challenging scenarios, including significant occlusions and object motions. Comprehensive experiments given two public benchmarks demonstrate that the proposed approach significantly outperforms state-of-art, category-level 6D tracking or dynamic SLAM methods. When compared against state-of-art methods that rely on an object instance CAD model, comparable performance is achieved, despite the proposed method’s reduced information requirements. An efficient implementation in CUDA provides a real-time performance of 10Hz for the entire framework. Code is available at: https://github.com/wenbowen123/BundleTrack

Learning a robot motor skill from scratch is impractically slow; so much so that in practice, learning must typically be bootstrapped using human demonstration. However, relying on human demonstration necessarily degrades the autonomy of robots that must learn a wide variety of skills over their operational lifetimes. We propose using kinematic motion planning as a completely autonomous, sample efficient way to bootstrap motor skill learning for object manipulation. We demonstrate the use of motion planners to bootstrap motor skills in two complex object manipulation scenarios with different policy representations: opening a drawer with a dynamic movement primitive representation, and closing a microwave door with a deep neural network policy. We also show how our method can bootstrap a motor skill for the challenging dynamic task of learning to hit a ball off a tee, where a kinematic plan based on treating the scene as static is insufficient to solve the task, but sufficient to bootstrap a more dynamic policy. In all three cases, our method is competitive with human-demonstrated initialization, and significantly out-performs starting with a random policy. This approach enables robots to to efficiently and autonomously learn motor policies for dynamic tasks without human demonstration.