Search

Relative localization is a fundamental requirement for the coordination of multiple robots. To date, existing research in relative localization mainly depends on the extraction of low-level geometry features such as planes, lines, and points, which may fail in challenging cases when initial error is large and overlapping area is low. In this paper, a novel approach named collaborative semantic map matching (COSEM) is proposed to estimate the relative transformation between robots. COSEM jointly performs multimodal information fusion, semantic data association, and optimization in a unified framework. Firstly, each robot applies a multimodal information fusion model to generate local semantic maps. Since the correspondences between local maps are latent variables, a flexible semantic data association strategy is proposed using Expectation-Maximization. Instead of assigning hard geometry data association, semantic association and geometry association are jointly estimated. Then, the minimization of the expected cost results in a rigid transformation matrix between two semantic maps. Evaluations on Semantic KITTI benchmarks and real world experiments show the improved accuracy, convergence, and robustness.

Recently, intelligent unmanned aerial vehicles (UAVs) have shown great advantages of flexibility and productivity in power line inspection, wherein robust detection of power lines from aerial images for automatic power line following navigation is required. However, identifying power lines accurately from a cluttered background is challenging due to the limited resolution of onboard cameras and the noisy environment. In this letter, we propose a novel power line detection method, denoted by CODNet, for the application of UAV navigation. Unlike existing works, the proposed method can extract features of power lines from cluttered backgrounds automatically and predict centers and orientations of power lines in the scene simultaneously. Besides, we introduce a new clustering method to summarize the average location and orientation of detected power lines as a guide for the automatic navigation of UAVs. Finally, experimental results demonstrate both the effectiveness and the superiority of the CODNet. This work was supported in part by the Natural Science Fund of China (NSFC) under Grant 51575186; in part by the Shanghai Science and Technology Action Plan under Grant 18DZ1204000, Grant 18510745500, Grant 18510750100, and Grant 18510730600; and in part by the Shanghai Aerospace Science and Technology Innovation Fund (SAST) under Grant 2019-080 and Grant 2019-116.

This article investigates the target docking problem for nonholonomic mobile robots using relative position and orientation measurements of the target. Considering the fact that target docking operations generally happen in unstructured environments where obstacles are left lying around, the whole docking process is divided into two phases, namely the approaching phase and the autonomous docking phase. In the approaching phase, the robot navigates in the operational environment until it reaches a switching region around the target, which triggers the autonomous docking phase. To account for the noise-corrupted and intermittent target observations, an extended Kalman filter-based relative pose estimation algorithm is designed to estimate the position and orientation of the target in the local reference frame of the robot. To achieve a perfect docking operation in the autonomous docking phase, which requires a zero-impact angle and a zero-lateral offset at the final engagement stage, a practical solution is proposed where motion planning and tracking control of the robot are incorporated into a unified scheme. Unlike most of the existing docking approaches in the literature which only account for static targets, our proposed strategy addresses the moving target docking problem as well. Real robot experiments have been performed to demonstrate the effectiveness of the proposed method.

Repetitive or ambiguous environment, where structures are highly similar and distinct geometric features are not sufficient, is one of the critical and challenging scenarios for mobile robots to perform global localization or simultaneous localization and mapping (SLAM) tasks. The robots are easy to get lost or mismatched to wrong places in such environments. The existing solutions either rely heavily on pre-installed infrastructures which are inflexible and expensive or rely on single sensor-based global localization methods whose initialization module is not capable enough to provide distinctive information. Thus, this article proposes a hierarchical probabilistic framework that addresses the problem of infrastructure-free mobile robot global localization in GPS-challenged repetitive environments by leveraging both the measured magnetic field (MF) and Light Detection and Ranging (LiDAR) information. The proposed hierarchical system mainly consists of: 1) coarse localization: MF-based localization and 2) fine localization: LiDAR-based localization (LBL). The ambient MF functions as a substitute for the GPS signal, which is considered not available or severely challenged in indoor/semi-indoor local environments. Based on the pre-built MF database, multiple coarse candidate robot poses can first be determined by the proposed initial pose estimation algorithm, which incorporates multivariate Gaussian observation model and random sample consensus (RANSAC) algorithm. Then, using the obtained multiple candidate poses as initialization, the robot can be localized more accurately by LiDAR-based fine localization. Extensive real-world experiments demonstrate that the proposed system achieves over 95% localization success rate and takes less than 2.0-m average traveling distance to localize the mobile robot.

This article proposes a modeling method for the fully closed-loop system of hemispherical resonator gyroscope (HRG) under force-to-rebalance mode. A fully closed-loop system consists of two parts: driving loop and detecting loop. First, the dynamic equations of the driving loop are derived to obtain the vibration differential equations of a hemispherical resonator driven by the electrostatic forces, to further obtain accurate output models of the wave amplitude and angular rate, which establishes a model of the driving loop. Next, the output model of the wave amplitude is analyzed to clarify the relationship between the amplitude voltage and wave amplitude. Simultaneously, the influences of interference factors on the amplitude voltage and angular rate voltage are analyzed, which can result in the HRG output error. Then, a model of detecting capacitance is built based on the deformation equations of the hemispherical resonator. A model of the detecting loop is established to analyze its performance and precision. Finally, by tests of amplitude voltage and resonant frequency, the accuracy of the models of the driving loop and detecting loop is proven. A model of the fully closed-loop system of the HRG is proposed based on the models of the driving loop and detecting loop. The HRG output errors caused by the resonator’s deformation, detecting error, and nonuniformity of the quality factor are identified and compensated accurately by the proposed model. It is verified by simulations and tests that the proposed model of the fully closed-loop system is effective, which can not only identify the main errors of the HRG accurately but also provide an accurate model for error analysis.

This article investigates the problem of distributed fixed-time attitude consensus control for multiple spacecraft when only a subset of spacecraft has access to a common reference attitude and angular velocity. Two fixed-time sliding-mode observers are proposed to estimate the reference attitude and angular velocity in fixed time, respectively. Based on estimates of the reference attitude and angular velocity, two distributed fixed-time attitude consensus control schemes are proposed under the mild assumption that there exists a path from the virtual leader to any follower and the communication topology among followers is undirected. In the first control scheme, the measurement of angular velocity is required. However, this requirement is removed in the second control scheme by using a fixed-time state observer to estimate the angular velocity of spacecraft in fixed time. Only a subset of spacecraft is required to access a common reference attitude and angular velocity in the two proposed control schemes. The stability and convergence of the resulting closed-loop systems are guaranteed by the Lyapunov approach. Finally, simulation results validate the effectiveness of the proposed control schemes.

This brief investigates the flexible leader-follower formation tracking problem for a group of nonholonomic mobile robots, while most of the formation control related work in the literature focuses on the rigid formation. The flexible formation discussed in this brief is defined in curvilinear coordinates in terms of longitudinal separations between robots along the reference trajectory and lateral deviations with respect to this trajectory. Unlike the previous studies on flexible formation control, this brief is under a more challenging assumption that the global position and orientation measurements are not available. To obtain the relative pose relationships amongst robots, a stereo camera is mounted on each follower. In consideration of the fact that visual observations are noise-corrupted and intermittently available, a particle filter-based relative pose estimation approach is employed to estimate the position and orientation of the leader in the local reference frame of the follower using the polluted and discontinuous information. Also, to form a flexible formation, the leader historical trajectory is reconstructed with respect to the current local frame attached on the follower, based on which a reference point is generated. In addition, this brief considers the situation where robots operate in unknown obstacle environments. To ensure robot safety in such environments, a multiobjective control law is proposed to balance reference tracking and collision avoidance in different situations. Simulation and real-robot experiment have been performed to demonstrate the efficacy of the proposed method. Accepted version

B cell dysfunction due to obesity can be associated with alterations in the levels of micro-RNAs (miRNAs). However, the role of miRNAs in these processes remains elusive. Here, we show that miR-802 is increased in the pancreatic islets of obese mouse models and demonstrate that inducible transgenic overexpression of miR-802 in mice causes impaired insulin transcription and secretion. We identify Foxo1 as a transcription factor of miR-802 promoting its transcription, and NeuroD1 and Fzd5 as targets of miR-802-dependent silencing. Repression of NeuroD1 in β cell and primary islets impairs insulin transcription and reduction of Fzd5 in β cell, which, in turn, impairs Ca2+ signaling, thereby repressing calcium influx and decreasing insulin secretion. We functionally create a novel network between obesity and β cell dysfunction via miR-802 regulation. Elucidation of the impact of obesity on microRNA expression can broaden our understanding of pathophysiological development of diabetes., Obesity predisposes to type 2 diabetes, but the mechanisms of obesity-associated β cell dysfunction are incompletely understood. Here the authors report that obesity increases the levels of miR-802, which impairs insulin transcription and secretion by targeting NeuroD1 and Fzd5.

Operating multiple robots in an unstructured environment is challenging due to its high complexity and uncertainty. In such applications, the integration of individual maps generated by heterogeneous sensors is a critical problem, especially the fusion of sparse and dense maps. This paper proposes a general multi-level probabilistic framework to address the integrated map fusion problem, which is independent of sensor type and SLAM algorithm employed. The key novelty of this paper is the mathematical formulation of the overall map fusion problem and the derivation of its probabilistic decomposition. The framework provides a theoretical basis for computing the relative transformations amongst robots and merging probabilistic map information. Since the maps generated by heterogeneous sensors have different physical properties, an expectation-maximization-based map-matching algorithm is proposed which automatically determines the number of voxels to be associated. Then, a time-sequential map merging strategy is developed to generate a globally consistent map. The proposed approach is evaluated in various environments with heterogeneous sensors, which demonstrates its accuracy and versatility in 3-D multirobot map fusion missions. Accepted version

This paper designs an active fault-tolerant control system for spacecraft attitude control in the presence of actuator faults, fault estimation errors, and control input constraints. The developed fault-tolerant control system is able to detect the actuator fault without false alarms caused by external disturbances, and also estimate the total fault effects accurately through an indirect fault identification approach, in which an auxiliary variable is utilized to build the relation between fault and system states. Once the fault identification is completed with certain degree of reconstruction accuracy, a fault-tolerant backstepping controller using the nonlinear virtual control input is reconfigured to accommodate the detected actuator faults effectively, in spite of actuator saturation limitations and fault estimation errors. Numerical simulation is carried out to demonstrate that the proposed active fault-tolerant control system is successful in fault detection, identification, and controller reconfiguration for handling actuator faults in attitude control systems.

Large-scale visual place recognition (VPR) is inherently challenging because not all visual cues in the image are beneficial to the task. In order to highlight the task-relevant visual cues in the feature embedding, the existing attention mechanisms are either based on artificial rules or trained in a thorough data-driven manner. To fill the gap between the two types, we propose a novel Semantic Reinforced Attention Learning Network (SRALNet), in which the inferred attention can benefit from both semantic priors and data-driven fine-tuning. The contribution lies in two-folds. (1) To suppress misleading local features, an interpretable local weighting scheme is proposed based on hierarchical feature distribution. (2) By exploiting the interpretability of the local weighting scheme, a semantic constrained initialization is proposed so that the local attention can be reinforced by semantic priors. Experiments demonstrate that our method outperforms state-of-the-art techniques on city-scale VPR benchmark datasets.

This article studies the distributed prescribed-time coordination problem for the attitude system of multiple spacecraft subject to actuator saturation under the communication topology containing a directed spanning tree. The prescribed-time coordination requires that every spacecraft in a formation reaches an agreement state in a preset time. This preset time is set by the user and independent of any other control parameters and the initial state of spacecraft. To achieve this goal, we first design a distributed prescribed-time observer for every follower spacecraft to estimate the state of leader spacecraft. Next, a prescribed-time attitude control law is proposed by utilizing the estimated state of the leader spacecraft. In the proposed control law, an adaptive compensation law is used to compensate the input saturation. Furthermore, the prescribed-time stability of multispacecraft system subject to input saturation under the proposed scheme is analyzed in theory and simulation. The simulation results show the high reliability and effectiveness of the scheme. This work was supported by the National Natural Science Foundation of China under Grant 61873312.

Prognosis of failing components under intermittent faults is challenging since intermittent faults gradually deteriorate in duration and magnitude over time meanwhile show the stochasticity of fault appearance and disappearance. To address the problem, this paper proposes an intelligent prognosis method for intermittently faulty components in electric scooter based on dual degradation processes. Firstly, fault detection and isolation is used to identify discrete faults and isolate possible intermittent faults in continuous components, where the fault isolation performance under multiple faults condition is improved by developing an extended fault signature matrix. Secondly, the adaptive competitive swarm optimization is proposed to identify the magnitude, appearing and disappearing instants of each intermittent fault for faulty components. After that, the dual degradation processes are established with the aid of tumbling window (TW), where the duration degradation process describes the evolutionary trend of the ratio of fault duration to the length of duration-TW, while the magnitude degradation process captures the evolutionary trend of maximum feature in magnitude-TW. With dual degradation processes and predefined failure thresholds, the remaining useful life of the faulty component is jointly predicted, where the degradation speed difference between duration and magnitude is considered. Finally, the proposed methods are validated by experiment results. This work was supported by the National Natural Science Foundation of China under Grants 62173119 and 61673154.

It is essential in many robot tasks to retrieve depth information, while it still remains a challenging problem to get robust depth in unfavorable conditions such as night or rainy environments. With the development of convolutional neural networks (CNNs), a large number of algorithms have emerged to tackle the problem of dark image enhancement and depth estimation, but there are few works focus on recovering depth map in dark environments and normal light condition. To meet this demand, we proposed a neural network which takes the paired stereo images in all light conditions as input and estimates the fully scaled depth map. The network contains a novel feature extractor and a stereo matching module which follows a light-weight manner to guarantee this work practical for real robotic applications. We introduced the Retinex Theory into depth estimation and trained the decomposition module with LOL dataset. Then it is adapted into depth estimation by fusing the decompose module into stereo matching algorithm. The whole network is then trained in an end-to-end manner. To demonstrate the robustness and effectiveness of our proposed method, we perform various studies and compare our results to the state-of-the-art algorithms in depth estimation as well as direct combination of image enhancement and stereo matching algorithm. We also collect stereo images in real night environments and present the improved performance of our network. Accepted version

Traffic forecasting is of great importance to vehicle routing, traffic signal control and urban planning. However, traffic forecasting task is challenging due to several factors, such as complex spatial topological structure and dynamic changing of traffic status. Most existing methods have limited ability to capture both spatial and temporal dependence of traffic data. In this paper, we propose a novel end-to-end deep learning model, Attention based Graph Bi-LSTM networks (AGBN) to perform the traffic forecasting task. It uses graph convolutional network (GCN) to extract spatial features and bidirectional long short-term memory networks (Bi-LSTM) to capture the temporal dependence. The attention mechanism is used to select relevant features at all time steps. Experiments show that our model could extract both spatial and temporal dependence well and outperforms other baselines on real-world traffic datasets. National Research Foundation (NRF) Accepted version This research is supported by National Research Foundation (NRF) Singapore, ST Engineering-NTU Corporate Lab under its NRF Corporate Lab@ University Scheme.

This paper proposes a hybrid condition monitoring approach, which integrates bond graph model-based diagnostic technique and data-driven remaining useful life (RUL) prediction, for a nonlinear mechatronic system. In this approach, various degrading faults can be considered and the physical degradation model is not required for RUL prediction. Firstly, an integrated fault signature matrix is proposed by the causal path of bicausal-bond graph model to improve fault isolation performance. After that, a biogeography-based optimization (BBO)-particle filter is developed for fault identification. For prognosis, an optimized extreme learning machine (OELM) is proposed where the hidden layer biases and input weights are optimized by BBO. The fault identification results provide data set to train the OELM for prognosis. Finally, the effectiveness of the approach is verified by simulation and experiment results.

Fusing 3-D maps generated by multiple robots in real/semi-real time distributed mapping systems are addressed in this paper. A 3-D occupancy grid-based approach for mapping is utilized to satisfy the real/semi-real time and distributed operating constraints. This paper proposes a novel hierarchical probabilistic fusion framework, which consists of uncertainty modeling, map matching, transformation evaluation, and map merging. Before the fusion of maps, the map features and their uncertainties are explicitly modeled and integrated. For map matching, a two-level probabilistic map matching (PMM) algorithm is developed to include high-level structural and low-level voxel features. In the PMM, the structural uncertainty is first used to generate a coarse matching between the maps and its result is then used to improve the voxel level map matching, resulting in a more efficient and accurate matching between maps with a larger convergence basin. The relative transformation output from PMM algorithm is then evaluated based on the Mahalanobis distance, and the relative entropy filter is used subsequently to integrate the map dissimilarities more accurately, completing the map fusion process. The proposed approach is evaluated using map data collected from both simulated and real environments, and the results validate the accuracy, efficiency, and the support for larger convergence basin of the proposed 3-D occupancy map fusion framework. Accepted version

This paper investigates the disturbance observer (DOB)-based neural adaptive control on the longitudinal dynamics of a flexible hypersonic flight vehicle (HFV) in the presence of wind effects. The coupling effect between flexible states and rigid body, and the accessional angle of attack (AOA) due to wind, is modeled as unknown disturbance, where the nonlinear DOB is constructed using the neural approximation. For the weight update in neural networks (NNs), a novel algorithm is proposed with the additional prediction error derived from the serial–parallel estimation model (SPEM) using both neural approximation and disturbance estimation. Different from previous work, the wind effect is taken into the hypersonic flight dynamics for realistic analysis, and the novel controller is designed using compound estimation, where the NN and the DOB are constructed to deal with aerodynamic uncertainty and unknown disturbance. Simulation studies of a flexible HFV with wind effects show that the proposed controller can achieve high tracking accuracy, while the compound estimation can closely follow the system uncertainty with fast convergence.

In conventional polishing processes, the polishing parameters are constant along the surface. Hence, if the desired material to be removed from the surface is not equally distributed, an over-polishing may occur for the areas with small material removal and under-polishing for the areas with large material removal. Consequently, the quality of the processed surface may not meet the manufacture requirements. In this paper, the authors proposed a polishing algorithm to deal with this problem using neural network (NNW) and genetic algorithm (GA). The NNW is used to predict the polishing performance parameters corresponding to a certain polishing parameters. In addition, the GA is employed to optimize the polishing parameters according to an objective function that includes the desired material removal and surface roughness improvement using the output from the trained NNW model. The effectiveness of the proposed algorithm is verified through experiments of polishing uneven surface.

In this paper, attitude coordinated tracking control algorithms for multiple spacecraft formation are investigated with consideration of parametric uncertainties, external disturbances, communication delays and actuator saturation. Initially, a sliding mode delay-dependent attitude coordinated controller is proposed under bounded external disturbances. However, neither inertia uncertainty nor actuator constraint has been taken into account. Then, a robust saturated delay-dependent attitude coordinated control law is further derived, where uncertainties and external disturbances are handled by Chebyshev neural networks (CNN). In addition, command filter technique is introduced to facilitate the backstepping design procedure, through which actuator saturation problem is solved. Thus the spacecraft in the formation are able to track the reference attitude trajectory even in the presence of time-varying communication delays. Rigorous analysis is presented by using Lyapunov-Krasovskii approach to demonstrate the stability of the closed-loop system under both control algorithms. Finally, the numerical examples are carried out to illustrate the efficiency of the theoretical results. Published version

In this paper, a model-based integrated framework for diagnosis and prognosis of electrical machines is proposed. Since the fault progression, compared to the dynamics of the measured terminal quantities, can be considered as a slow process, it can be approximately modeled by the step changes of certain model-based parameters, representing fault severity levels. Diagnosis process can reliably detect the fault occurrence and severity changes, from the incipient stage of fault development, by which the parameter estimation and prognosis process, leading to calculation of remaining useful life, can be promptly commenced. A generic method using multiple fault indicators for diagnosis and particle filter for prognosis is proposed for electrical machines under stator winding interturn fault. The proposed framework is experimentally validated using real machine signatures consisting of only stator voltage and current measurements.

This paper investigates a velocity-free finite-time attitude control scheme for a rigid spacecraft with actuator saturation and external disturbances. Initially, a finite-time observer is designed to compensate the unknown angular velocity information. With the estimated values, a finite-time controller is further developed under which the time-varying reference attitude trajectory can be tracked precisely. Moreover, input saturation problem is solved by adaptive method. Rigorous Lyapunov-based analysis shows that the states of the closed-loop system converge to a small neighborhood of the origin in finite time. Finally, the dynamic performance of attitude control system is presented by numerical simulation examples and the superiority of the finite-time control scheme is verified.

To promote operational intelligence, improve surface quality, and reduce manpower dependence, a novel high-bandwidth end-effector with active force control for robotic polishing was proposed. Using this end-effector as a mini robot, a macro-mini robot for polishing processing was constructed, in which the macro robot provides posture control during polishing operations, whereas the mini-robot provides constant force control. By minimizing the inertia along the spindle in this configuration, the end-effector obtains a force control bandwidth of 200 Hz. Through a series of comparative experiments with different contact forces and feed rates, the proposed design was proven to have a smaller overshoot, a faster response, and a shorter settling time than the conventional method based on macro robot (KUKA iiwa) controlled force. The roughness of the workpiece reached $0.4~\mu \text{m}$ after polishing with the macro-mini robot, indicating the efficiency of this end-effector in high-precision material removal and surface polishing operations. Published version

Performing collaborative semantic mapping is a critical challenge for cooperative robots to enhance their comprehensive contextual understanding of the surroundings. This paper bridges the gap between the advances in collaborative geometry mapping that relies on pure geometry information fusion, and single robot semantic mapping that focuses on integrating continuous raw sensor data. In this paper, a novel hierarchical collaborative probabilistic semantic mapping framework is proposed, where the problem is formulated in a distributed setting. The key novelty of this work is the modelling of the hierarchical semantic map fusion framework and its mathematical derivation of its probability decomposition. At the single robot level, the semantic point cloud is obtained by combining information from heterogeneous sensors and used to generate local semantic maps. At the collaborative robots level, local maps are shared among robots for global semantic map fusion. Since the voxel correspondence is unknown between local maps, an Expectation-Maximization approach is proposed to estimate the hidden data association. Then, Bayesian rule is applied to perform semantic and occupancy probability update. Experimental results on the UAV (Unmanned Aerial Vehicle) and the UGV (Unmanned Ground Vehicle) platforms show the high quality of global semantic maps, demonstrating the accuracy andutility in practical missions. National Research Foundation (NRF) Accepted version This work was supported by National Research Foundation Singapore, ST Engineering-NTU Corporate Lab under its NRF Corporate Lab@ University Scheme.

Repetitive environment is a challenging scenario for mobile robot global localization due to its highly similar structures and lack of distinctive features. Existing solutions in such environments rely heavily on pre-installed infrastructures, which are neither flexible nor cost-effective. Besides, few of the previous research have been focused on the implementation of infrastructure-free localization approaches in repetitive scenarios. Thus, this paper serves as a survey to investigate the problem of infrastructure-free mobile robot global localization with low-cost and efficient sensors in repetitive environments. Three of the most popular infrastructure-free localization methods, namely LiDAR-based localization (LBL), vision-based localization (VBL), and magnetic field-based localization (MFL), are analyzed and evaluated. Extensive global localization experiments are conducted in real-world repetitive scenarios and the results demonstrate that VBL methods perform slightly better than LBL and MFL methods. The overall evaluations indicate that infrastructure-free global localization in repetitive environment is still a challenging problem which deserves more research efforts to develop new solutions. National Research Foundation (NRF) Accepted version

Performing collaborative semantic mapping is a critical challenge for cooperative robots to maintain a comprehensive contextual understanding of the surroundings. Most of the existing work either focus on single robot semantic mapping or collaborative geometry mapping. In this paper, a novel hierarchical collaborative probabilistic semantic mapping framework is proposed, where the problem is formulated in a distributed setting. The key novelty of this work is the mathematical modeling of the overall collaborative semantic mapping problem and the derivation of its probability decomposition. In the single robot level, the semantic point cloud is obtained based on heterogeneous sensor fusion model and is used to generate local semantic maps. Since the voxel correspondence is unknown in collaborative robots level, an Expectation-Maximization approach is proposed to estimate the hidden data association, where Bayesian rule is applied to perform semantic and occupancy probability update. The experimental results show the high quality global semantic map, demonstrating the accuracy and utility of 3D semantic map fusion algorithm in real missions.

Enabling long-term operation during day and night for collaborative robots requires a comprehensive understanding of the unstructured environment. Besides, in the dynamic environment, robots must be able to recognize dynamic objects and collaboratively build a global map. This paper proposes a novel approach for dynamic collaborative mapping based on multimodal environmental perception. For each mission, robots first apply heterogeneous sensor fusion model to detect humans and separate them to acquire static observations. Then, the collaborative mapping is performed to estimate the relative position between robots and local 3D maps are integrated into a globally consistent 3D map. The experiment is conducted in the day and night rainforest with moving people. The results show the accuracy, robustness, and versatility in 3D map fusion missions.

Reinforcement learning has obtained increasing attention in mobile robot mapless navigation in recent years. However, there are still some obvious challenges including the sample efficiency, safety due to dilemma of exploration and exploitation. These problems are addressed in this paper by proposing the Human-in-Loop Policy Search (HILPS) framework, where learning from demonstration, learning from human intervention and Near Optimal Policy strategies are integrated together. Firstly, the former two make sure that expert experience grant mobile robot a more informative and correct decision for accomplishing the task and also maintaining the safety of the mobile robot due to the priority of human control. Then the Near Optimal Policy (NOP) provides a way to selectively store the similar experience with respect to the preexisting human demonstration, in which case the sample efficiency can be improved by eliminating exclusively exploratory behaviors. To verify the performance of the algorithm, the mobile robot navigation experiments are extensively conducted in simulation and real world. Results show that HILPS can improve sample efficiency and safety in comparison to state-of-art reinforcement learning. Accepted version

The problem of attitude tracking control for spacecraft with limited communication rate is addressed in this article. To reduce the communication burden, an adaptive event-triggered control scheme is proposed. In the control scheme, only the sampling states at the event-triggering instants are sent to the control module, which can considerably decrease the data transmission rate. To address the inertia uncertainties and external disturbances, a radial basis function neural network (NN) is introduced. The bound of the uncertainties and disturbances is estimated for the proposed control scheme, which can simplify the NN and reduce the computation. Since the event-triggered error signal is discontinuous due to the event-triggered mechanism, the closed-loop system is formulated as an impulsive dynamical system to obtain the stability properties of the system. Finally, simulation results are given to demonstrate the effectiveness of the proposed control scheme.

Accurate estimation of extrinsic parameter (rotation matrix and translation vector) between heterogeneous sensors is important for fusing complementary information. However, the extrinsic calibration between a sparse 3D LiDAR and a thermal camera is challenging, mainly because of the difficulties to accurately extract common features from a sparse point cloud and a thermal image which has limited-FOV and low-resolution. Previous methods either rely on a dense depth sensor or a visual camera to facilitate the feature extraction process. To address the problem, SLAT-Calib (Sparse Lidar And Thermal camera Calibration) is proposed. By observing that circular holes could be detected from both sensors, a specially designed calibration board (a rectangular board with four circular holes) is introduced. Four circle centers in 3D space are used as common features. The benefit is point features are accurate and reliable for feature matching. To extract four circle centers from the thermal camera, three steps are carried out: First, a method is proposed to accurately detect the four circles. Then, the homography matrix of the calibration board can be figured out. Lastly, 3D coordinates of the circle centers are calculated by decomposing the homography matrix. From the LiDAR frame, the four circle centers can be segmented out as long as two laser beams pass through each circle. At last, optimal extrinsic parameter is calculated by minimizing the matching error between the four pairs of 3D circle centers. Quantitative and qualitative experiments are carried out. In simulation, SLAT-Calib outperforms two methods by a large margin. In real environment, it achieves a re-projection error (RMSE) of 0.62 pixel.

Ensuring safety of both traffic participants and passengers is an important challenge for rapidly growing autonomous vehicle technology. To this purpose, intelligent vehicles not only have to drive safe but must be able to safeguard itself from other abnormally driving vehicles and avoid potential collisions. Anomaly detection is one of the essential abilities in behavior analysis, which can be used to infer the moving intention of other vehicles and provide evidence for collision risk assessment. In this paper, we propose a behavior analysis method based on Hidden Markov Model (HMM) to assess the driving behavior of vehicles on the road and detect anomalous moments. The algorithm uses the real-time velocity and position of the surrounding vehicles provided by the Conditional Monte Carlo Dense Occupancy Tracker (CMCDOT) framework. Next, by associating with the road information, the movement of each vehicle can be classified into several observation states, namely, Approaching, Braking, Lane Changing, and Lane Keeping. Finally, by chaining these observation states using a Markov model, the abnormality of driving behavior can be inferred into Normal, Attention, and Risk. We perform experiments using CARLA simulator environment to simulate abnormal driving behaviors, and we provide results showing the successful detection of abnormal situations.

Place recognition plays a vital role in eliminating accumulated drift from visual odometry in SLAM system. Bag- of-Words (BoW) -based approach is the most popular solution due to its efficiency and robustness. We propose to use Line- Junction-Line (LJL) to build a BoW for place recognition in urban environments. LJL is a simple structure of two lines with their intersection. Different from point features which are detected based on pixel intensity patterns, it represents structure with physical existence, which is more robust to challenging scenarios. Moreover, its descriptor is distinctive and encodes the relationship between the two lines. Experiments on KITTI dataset show the effectiveness of the proposed method compared to loop detection using BoW trained with either point or line features. NRF (Natl Research Foundation, S’pore) Accepted version

Performing collaborative semantic mapping is a critical challenge for multi-robot systems to maintain a comprehensive contextual understanding of the surroundings. In this paper, a novel hierarchical semantic map fusion framework is proposed, where the problem is addressed in low-level single robot semantic mapping and high level global semantic map fusion. In the single robot semantic mapping process, Bayesian rule is used for label fusion and occupancy probability updating, where the semantic information is added to the geometric map grid. High level global semantic map fusion covers decentralized map sharing and global semantic map updating. Collaborative semantic mapping is conducted in two scenarios, that is, NTU dataset and the KITTI dataset. The results show the high quality of the global semantic map, which demonstrates the utility and versatility of 3D semantic map fusion algorithm. Accepted version