IROS 2020 论文列表

While Lidar beams are often represented as rays, they actually have finite beam width and this width impacts the measured shape and size of objects in the scene. Here we investigate the effects of beam width on measurements of thin objects such as vertical poles. We propose a model for beam divergence and show how this can explain both object dilation and erosion. We develop a calibration method to estimate beam divergence angle. This calibration method uses one or more vertical poles observed from a Lidar on a moving platform. In addition, we derive an incremental method for using the calibrated beam angle to obtain accurate estimates of thin object diameters, observed from a Lidar on a moving platform. Our method achieves significantly more accurate diameter estimates than is obtained when beam divergence is ignored.

This paper proposes a 3D LiDAR simultaneous localization and mapping (SLAM) method that improves accuracy, robustness, and computational efficiency for an iterative closest point (ICP) algorithm employing a locally approximated geometry with clusters of normal distributions. In comparison with previous normal distribution-based ICP methods, such as normal distribution transformation and generalized ICP, our ICP method is simply stabilized with normalization of the cost function by the Frobenius norm and a regularized covariance matrix. The previous methods are stabilized with principal component analysis, whose computational cost is higher than that of our method. Moreover, our SLAM method can reduce the effect of incorrect loop closure constraints. The experimental results show that our SLAM method has advantages over open source state-of-the-art methods, including LOAM, LeGO-LOAM, and hdl_graph_slam.

Detecting small obstacles on the road is critical for autonomous driving. In this paper, we present a method to reliably detect such obstacles through a multi-modal framework of sparse LiDAR(VLP-16) and Monocular vision. LiDAR is employed to provide additional context in the form of confidence maps to monocular segmentation networks. We show significant performance gains when the context is fed as an additional input to monocular semantic segmentation frameworks. We further present a new semantic segmentation dataset to the community, comprising of over 3000 image frames with corresponding LiDAR observations. The images come with pixel-wise annotations of three classes off-road, road, and small obstacle. We stress that precise calibration between LiDAR and camera is crucial for this task and thus propose a novel Hausdorff distance based calibration refinement method over extrinsic parameters. As a first benchmark over this dataset, we report our results with 73 % instance detection up to a distance of 50 meters on challenging scenarios. Qualitatively by showcasing accurate segmentation of obstacles less than 15 cms at 50m depth and quantitatively through favourable comparisons vis a vis prior art, we vindicate the method's efficacy. Our project and dataset is hosted at https://small-obstacle-dataset.github.io/.

Truly autonomous driving without the need for human intervention can only be attained when self-driving cars fully understand their surroundings. Most of these vehicles rely on a suite of active and passive sensors. LiDAR sensors are a cornerstone in most of these hardware stacks, and leveraging them as a complement to other passive sensors such as RGB cameras is an enticing goal. Understanding the semantic class of each point in a LiDAR sweep is important, as well as knowing to which instance of that class it belongs to. To this end, we present a novel, single-stage, and real-time capable panoptic segmentation approach using a shared encoder with a semantic and instance decoder. We leverage the geometric information of the LiDAR scan to perform a novel, distance- aware tri-linear upsampling, which allows our approach to use larger output strides than using transpose convolutions leading to substantial savings in computation time. Our experimental evaluation and ablation studies for each module show that combining our geometric and semantic embeddings with our learned, variable instance thresholds, a category-specific loss, and the novel trilinear upsampling module leads to higher panoptic quality. We will release the code of our approach in our LiDAR processing library LiDAR-Bonnetal [27].

In this paper, a global descriptor for a LiDAR point cloud, called LiDAR Iris, is proposed for fast and accurate loop-closure detection. A binary signature image can be obtained for each point cloud after several LoG-Gabor filtering and thresholding operations on the LiDAR-Iris image representation. Given two point clouds, their similarities can be calculated as the Hamming distance of two corresponding binary signature images extracted from the two point clouds, respectively. Our LiDAR-Iris method can achieve a pose-invariant loop-closure detection at a descriptor level with the Fourier transform of the LiDAR-Iris representation if assuming a 3D (x,y,yaw) pose space, although our method can generally be applied to a 6D pose space by re-aligning point clouds with an additional IMU sensor. Experimental results on five road-scene sequences demonstrate its excellent performance in loop-closure detection.

Dynamic obstacle avoidance is one crucial component for compliant navigation in crowded environments. In this paper we present a system for accurate and reliable detection and tracking of dynamic objects using noisy point cloud data generated by stereo cameras. Our solution is real-time capable and specifically designed for the deployment on computationally-constrained unmanned ground vehicles. The proposed approach identifies individual objects in the robot's surroundings and classifies them as either static or dynamic. The dynamic objects are labeled as either a person or a generic dynamic object. We then estimate their velocities to generate a 2D occupancy grid that is suitable for performing obstacle avoidance. We evaluate the system in indoor and outdoor scenarios and achieve real-time performance on a consumergrade computer. On our test-dataset, we reach a MOTP of 0.07 ± 0.07m, and a MOTA of 85.3% for the detection and tracking of dynamic objects. We reach a precision of 96.9% for the detection of static objects.

With monocular Visual-Inertial Odometry (VIO) system, 3D point cloud and camera motion can be estimated simultaneously. Because pure sparse 3D points provide a structureless representation of the environment, generating 3D mesh from sparse points can further model the environment topology and produce dense mapping. To improve the accuracy of 3D mesh generation and localization, we propose a tightly-coupled monocular VIO system, PLP-VIO, which exploits point features and line features as well as plane regularities. The co-planarity constraints are used to leverage additional structure information for the more accurate estimation of 3D points and spatial lines in state estimator. To detect plane and 3D mesh robustly, we combine both the line features with point features in the detection method. The effectiveness of the proposed method is verified on both synthetic data and public datasets and is compared with other state-of-the-art algorithms.

Learning is usually performed by observing real robot executions. Physics-based simulators are a good alternative for providing highly valuable information while avoiding costly and potentially destructive robot executions. We present a novel approach for learning the probabilities of symbolic robot action outcomes. This is done leveraging different environments, such as physics-based simulators, in execution time. To this end, we propose MENID (Multiple Environment Noise Indeterministic Deictic) rules, a novel representation able to cope with the inherent uncertainties present in robotic tasks. MENID rules explicitly represent each possible outcomes of an action, keep memory of the source of the experience, and maintain the probability of success of each outcome. We also introduce an algorithm to distribute actions among environments, based on previous experiences and expected gain. Before using physics-based simulations, we propose a methodology for evaluating different simulation settings and determining the least time-consuming model that could be used while still producing coherent results. We demonstrate the validity of the approach in a dismantling use case, using a simulation with reduced quality as simulated system, and a simulation with full resolution where we add noise to the trajectories and some physical parameters as a representation of the real system.

Multi-functional cells with cooperating teams of robots promise to be flexible, robust, and efficient and, thus, are a key to future factories. However, their programming is tedious and AI-based planning for multiple robots is computationally expensive. In this work, we present a modular and efficient two-layer planning approach for multi-robot assembly. The goal is to generate the program for coordinated teams of robots from an (enriched) 3D model of the target assembly. Although the approach is both motivated and evaluated with LEGO, which is a challenging variant of blocks world, the approach can be customized to different kinds of assembly domains.

In this paper, an approach for automatic peg-in-hole assembly is proposed. The task is divided into two main steps: searching phase and inserting phase. First, a multilayer perceptron network is designed to address the hole search problem and a hybrid force position controller is introduced to ensure a safe and stable interaction with the external environment. Then, for the inserting phase, a variable impedance controller is adopted based on the fuzzy Q-learning algorithm to yield compliant behavior from the robot during the hole insertion process. This approach is a practical and general approach to solve complex peg-in-hole assembly problems by taking advantage of both learning-based algorithms and force control strategies, which can greatly improve the efficiency and safety of the industrial manufacturing process without identifying the unknown contact model and tuning tedious parameters. Finally, the peg-in-hole experimental results for an industrial robot verified the effectiveness and robustness of the proposed approach.

Safety is a critical issue in learning-based robotic and autonomous systems as learned information about their environments is often unreliable and inaccurate. In this paper, we propose a risk-aware motion control tool that is robust against errors in learned distributional information about obstacles moving with unknown dynamics. The salient feature of our model predictive control (MPC) method is its capability of limiting the risk of unsafety even when the true distribution deviates from the distribution estimated by Gaussian process (GP) regression, within an ambiguity set. Unfortunately, the distributionally robust MPC problem with GP is intractable because the worst-case risk constraint involves an infinite-dimensional optimization problem over the ambiguity set. To remove the infinite-dimensionality issue, we develop a systematic reformulation approach exploiting modern distributionally robust optimization techniques. The performance and utility of our method are demonstrated through simulations using a nonlinear car-like vehicle model for autonomous driving.

Dynamic control for robotic automation tasks is traditionally designed and optimized with a model-based approach, and the performance relies heavily upon accurate system modeling. However, modeling the true dynamics of increasingly complex robotic systems is an extremely challenging task and it often renders the automation system to operate in a non-optimal condition. Notably, many industrial robotic applications involve repetitive motions and constantly generate a large amount of motion data under the non-optimal condition. These motion data contain rich information, and therefore an intelligent automation system should be able to learn from these non-optimal motion data to drive the system to operate optimally in a data-driven manner. In this paper, we propose a learning-based controller optimization algorithm for repetitive robotic tasks. To achieve this, a multi-objective cost function is designed to take into consideration both the trajectory tracking accuracy and smoothness, and then a data-driven approach is developed to estimate the gradient and Hessian based on the motion data for optimization without relying on the dynamic model. Experiments based on a magnetically-levitated nanopositioning system are conducted to demonstrate the effectiveness and practical appeals of the proposed algorithm in repetitive robotic automation tasks.

Manipulation and assembly tasks require non-trivial planning of actions depending on the environment and the final goal. Previous work in this domain often assembles particular instances of objects from known sets of primitives. In contrast, we aim to handle varying sets of primitives and to construct different objects of a shape category. Given a single object instance of a category, e.g. an arch, and a binary shape classifier, we learn a visual policy to assemble other instances of the same category. In particular, we propose a disassembly procedure and learn a state policy that discovers new object instances and their assembly plans in state space. We then render simulated states in the observation space and learn a heatmap representation to predict alternative actions from a given input image. To validate our approach, we first demonstrate its efficiency for building object categories in state space. We then show the success of our visual policies for building arches from different primitives. Moreover, we demonstrate (i) the reactive ability of our method to re-assemble objects using additional primitives and (ii) the robust performance of our policy for unseen primitives resembling building blocks used during training. Our visual assembly policies are trained with no real images and reach up to 95% success rate when evaluated on a real robot.

Heuristic-based graph search algorithms like A* are frequently used to solve motion planning problems in many domains. For most practical applications, it is infeasible and unnecessary to pre-compute the graph representing the whole search space. Instead, these algorithms generate the graph incrementally by applying a fixed set of actions (frequently called motion primitives) to find the successors of every node that they need to evaluate. In many domains, it is possible to define actions (called adaptive motion primitives) that are not pre-computed but generated on the fly. The generation and validation of these adaptive motion primitives is usually quite expensive compared to pre-computed motion primitives. However, they have been shown to drastically speed up search if used judiciously. In prior work, ad hoc techniques like fixed thresholds have been used to limit unsuccessful evaluations of these actions. In this paper, we propose a learning-based approach to make more intelligent decisions about when to evaluate them. We do a thorough empirical evaluation of our model on a 3 degree-of-freedom (dof) motion planning problem for navigation using the Reeds-Shepp path as an adaptive motion primitive. Our experiments show that using our approach in conjunction with search algorithms leads to over 2x speedup in planning time.

We present a robotic system capable of navigating autonomously by following a line and taking good quality pictures of people. When a group of people is detected, the robot rotates towards them and then back to line while continuously taking pictures from different angles. Each picture is processed in the cloud where its quality is estimated in a two-stage algorithm. First, features such as the face orientation and likelihood of facial emotions are input to a fully connected neural network to assign a quality score to each face. Second, a representation is extracted by abstracting faces from the image and it is input to a Convolutional Neural Network (CNN) to classify the quality of the overall picture. We collected a dataset in which a picture was labeled as good quality if subjects are well-positioned in the image and oriented towards the camera with a pleasant expression. Our approach detected the quality of pictures with 78.4% accuracy in this dataset and received a better mean user rating (3.71/5) than a heuristic method that uses photographic composition procedures in a study where 97 human judges rated each picture. Statistical analysis against the state-of-the-art verified the quality of the resulting pictures.

Convolutional Neural Networks (CNNs) have been successfully applied for relative camera pose estimation from labeled image-pair data, without requiring any handengineered features, camera intrinsic parameters or depth information. The trained CNN can be utilized for performing pose based visual servo control (PBVS). One of the ways to improve the quality of visual servo output is to improve the accuracy of the CNN for estimating the relative pose estimation. With a given state-of-the-art CNN for relative pose regression, how can we achieve an improved performance for visual servo control? In this paper, we explore switching of CNNs to improve the precision of visual servo control. The idea of switching a CNN is due to the fact that the dataset for training a relative camera pose regressor for visual servo control must contain variations in relative pose ranging from a very small scale to eventually a larger scale. We found that, training two different instances of the CNN, one for large-scale-displacements (LSD) and another for small-scale-displacements (SSD) and switching them during the visual servo execution yields better results than training a single CNN with the combined LSD+SSD data. However, it causes extra storage overhead and switching decision is taken by a manually set threshold which may not be optimal for all the scenes. To eliminate these drawbacks, we propose an efficient switching strategy based on model agnostic meta learning (MAML) algorithm. In this, a single model is trained to learn parameters which are simultaneously good for multiple tasks, namely a binary classification for switching decision, a 6DOF pose regression for LSD data and also a 6DOF pose regression for SSD data. The proposed approach performs far better than the naive approach, while storage and run-time overheads are almost negligible.

In this paper, we present a new locomotion control method for soft robot snakes. Inspired by biological snakes, our control architecture is composed of two key modules: A reinforcement learning (RL) module for achieving adaptive goal-tracking behaviors with changing goals, and a central pattern generator (CPG) system with Matsuoka oscillators for generating stable and diverse locomotion patterns. The two modules are interconnected into a closed-loop system: The RL module, analogizing the locomotion region located in the midbrain of vertebrate animals, regulates the input to the CPG system given state feedback from the robot. The output of the CPG system is then translated into pressure inputs to pneumatic actuators of the soft snake robot. Based on the fact that the oscillation frequency and wave amplitude of the Matsuoka oscillator can be independently controlled under different time scales, we further adapt the option-critic framework to improve the learning performance measured by optimality and data efficiency. The performance of the proposed controller is experimentally validated with both simulated and real soft snake robots.

Tactile sensing has been used for a variety of robotic exploration and manipulation tasks but a common constraint is a requirement for a large amount of training data. This paper addresses the issue of data-efficiency by proposing a novel method for online learning based on a Gaussian Process Latent Variable Model (GP-LVM), whereby the robot learns from tactile data whilst performing a contour following task thus enabling generalisation to a wide variety of tactile stimuli. The results show that contour following is successful with comparatively little data and is robust to novel stimuli. This work highlights that even with a simple learning architecture there are significant advantages to be gained in efficient and robust task performance by using latent variable models and online learning for tactile sensing tasks. This paves the way for a new generation of robust, fast, and data-efficient tactile systems.

Long-tail and rare event problems become crucial when autonomous driving algorithms are applied in the real world. For the purpose of evaluating systems in challenging settings, we propose a generative framework to create safety-critical scenarios for evaluating specific task algorithms. We first represent the traffic scenarios with a series of autoregressive building blocks and generate diverse scenarios by sampling from the joint distribution of these blocks. We then train the generative model as an agent (or a generator) to search the risky scenario parameters for a given driving algorithm. We treat the driving algorithm as an environment that returns high reward to the agent when a risky scenario is generated. The whole process is optimized by the policy gradient reinforcement learning method. Through the experiments conducted on several scenarios in the simulation, we demonstrate that the proposed framework generates safety-critical scenarios more efficiently than grid search or human design methods. Another advantage of this method is its adaptiveness to the routes and parameters.