Ships are a fundamental element of water transport traffic scenarios and the primary focus of waterway traffic monitoring. Shipping transportation, as a predominant mode of transportation, has witnessed rapid development in recent years. The automated classification of inland river ships serves as the foundation for the digitization and intelligent management of inland waterway transportation. It is crucial for facilitating the high-quality development of the shipping industry. The predominant approach for inland ship classification relies on visual sensors and synthetic aperture radar, which are limited in providing detailed 3D geometric information and are affected by varying weather and lighting conditions. In this paper, we propose a LiDAR-based ship classification method for inland waterways to address this issue. This method involves background filtering and target detection on the original point cloud, generating a dataset of point clouds of inland ships, and using PointNet to learn and classify ship point cloud features. Moreover, for the first time, we propose a point cloud classification framework for multi-channel feature fusion. The proposed framework fuses LiDAR ring ID, intensity, and geometric features into a unified point cloud representation. Based on the fused point cloud data, an improved model with a point-wise attention mechanism is employed for feature extraction and classification. Our method achieves an accuracy of 97.33%, surpassing the geometric information-only method by 2.94%. This result effectively demonstrates the method’s efficacy in extracting features and classifying LiDAR point cloud ships.

This article proposes a novel discounted inverse reinforcement learning (DIRL) algorithm for linear quadratic (LQ) control of unknown continuous-time (CT) systems with partially observable states and an unknown discounted value function. Existing DIRL methods predominantly rely on full-state feedback, limiting their applicability to practical scenarios where only input-output data are available. To this end, a state reconstruction method is designed for the system controlled by an expert using the measured desired output. Based on this, a model-free output-feedback (OPFB) DIRL algorithm is presented to iteratively solve the unknown value function and the corresponding optimal OPFB control policy equivalent to the expert control policy. The convergence of the proposed algorithm and the nonuniqueness of solutions are rigorously analyzed. Finally, comprehensive simulations reveal the effectiveness of the proposed algorithm in recovering the expert control policy and its superior computational efficiency compared to state-of-the-art (SOTA) methods.

Modality imbalance is a significant challenge for multi-modal interaction at various depths in multispectral pedestrian detection under varying illumination environments. To overcome the limitations of current cross attention in addressing the modality imbalance, we propose the Cross-Modal Dual-Stream Feature Interaction Transformer (CDFIT). CDFIT capitalizes on the Transformer’s ability to learn long-range dependencies, extracting global intra-modal and inter-modal correlations during the feature interaction phase. Crucially, in order to effectively eliminate the interference of the self-attention within one modality to the alternative one, we propose horizontal and vertical correlation decoupling modes to divide and reassemble the attention maps in CDFIT. This facilitates more purified inter-modal attention while preserving relevant intra-modal self-attention, reducing the information interference. Meanwhile, in CDFIT, we expand Transformer into dual-stream pathways to align and assemble the information from RGB and thermal modalities across depths separately, thereby greatly enhancing the performance of multispectral object detection. Comprehensive experiments and ablation studies on benchmark datasets demonstrate that CDFIT achieves superior performance compared with state-of-the-art methods.

Solving real-world complex tasks using reinforcement learning (RL) without high-fidelity simulation environments or large amounts of offline data can be quite challenging. Online RL agents trained in imperfect simulation environments can suffer from severe sim-to-real issues. Offline RL approaches although bypass the need for simulators, often pose demanding requirements on the size and quality of the offline datasets. The recently emerged hybrid offline-and-online RL provides an attractive framework that enables joint use of limited offline data and imperfect simulator for transferable policy learning. In this paper, we develop a new algorithm, called$\mathrm{H} 2 \mathrm{O}+$, which offers great flexibility to bridge various choices of offline and online learning methods, while also accounting for dynamics gaps between the real and simulation environments. Through extensive simulation and real-world robotics experiments, we demonstrate superior performance and flexibility of$\mathbf{H 2 O}+$over advanced cross-domain online and offline RL algorithms.

Serving as a critical component in laser pointing systems, fast steering mirrors (FSMs) encounter various control challenges in precise tracking tasks. In response to these challenges, a disturbance-rejection model predictive control (DR-MPC) method is proposed in this work. Initially, an auxiliary state space model of the tracking error is formulated by exploiting the minimal polynomials of the reference and disturbance signals. Subsequently, an implicitly constrained optimization problem is constructed, eliminating the need for explicit modeling of the unmeasurable disturbance. The minimal polynomial framework enables the transformation of the original problem into an explicit constrained optimization problem by linking past and predicted control inputs. Finally, the coordinate descent method is employed to iteratively solve the constrained optimization problem, enabling the successful deployment of the proposed controller on 10kHz high-speed hardware. Experimental results indicate that the proposed DR-MPC method exhibits substantial advantages regarding disturbance rejection, constraint handling, and dynamic tracking.

Developing controllers that generalize across diverse robot morphologies remains a significant challenge in legged locomotion. Traditional approaches either create specialized controllers for each morphology or compromise performance for generality. This paper introduces a two-stage teacher-student framework that bridges this gap through policy distillation. First, we train specialized teacher policies optimized for individual morphologies, capturing the unique optimal control strategies for each robot design. Then, we distill this specialized expertise into a single Transformer-based student policy capable of controlling robots with varying leg configurations. Our experiments across five distinct legged morphologies demonstrate that our approach preserves morphology-specific optimal behaviors, with the Transformer architecture achieving 94.47% of teacher performance on training morphologies and 72.64% on unseen robot designs. Comparative analysis reveals that Transformer-based architectures consistently outperform MLP baselines by leveraging attention mechanisms to effectively model joint relationships across different kinematic structures. We validate our approach through successful deployment on a physical quadruped robot, demonstrating the practical viability of our morphology-agnostic control framework. This work presents a scalable solution for developing universal legged robot controllers that maintain near-optimal performance while generalizing across diverse morphologies.

This paper proposes a novel Q-learning-based dual-loop force tracking control framework for robot grinding tasks in uncertain environments. A complete system state-space model is established, incorporating interaction dynamics and the desired force. By augmenting the system state, a discount cost function is defined to quantify the tracking errors of the force and reference trajectory. The modified Q-learning method is systematically designed to iteratively compute the optimal control gain in a model-free manner. To mitigate force overshoot during the transition from free space to contact space, a force reference model and a transition mechanism for the control gain are designed. Simulations and experiments validate the method’s effectiveness in precise force tracking with minimal overshoot and robustness to environmental variations.

Linear quadratic control with unknown value functions and dynamics is extremely challenging, and most of the existing studies have focused on the regulation problem, incapable of dealing with the tracking problem. To solve both linear quadratic regulation and tracking problems for continuous-time systems with unknown value functions, this article develops a discounted inverse reinforcement learning (DIRL) method that inherits the model-independent property of reinforcement learning (RL). More specifically, we first formulate a standard paradigm for solving linear quadratic control using DIRL. To recover the value function and the target control gain, an error metric is elaborately constructed, and a quasi-Newton algorithm is adopted to minimize it. Furthermore, three DIRL algorithms, including model-based, model-free off-policy, and model-free on-policy algorithms, are proposed. The latter two rely on the expert's demonstration data or the online observed data, requiring no prior knowledge of the system dynamics and value function. The stability, convergence, and existence conditions of multiple solutions are thoroughly analyzed. Finally, numerical simulations demonstrate the effectiveness of the theoretical results.

This work develops an output feedback adaptive tracking control method based on dynamic regressor extension and mixing (DREM) for discrete-time uncertain parameter systems. A piecewise DREM estimator is designed for the uncertain parameters under conditions strictly weaker than the persistently excited condition, exhibiting the ability to capture the actual system dynamics in finite time. Accurate parameter estimation guarantees the performance of the controller utilizing the DREM estimator. Then, an adaptive optimal controller for any given reference trajectory is designed within the framework of receding horizon control. The system state and control input are theoretically guaranteed to remain bounded during tracking. The adaptive controller is restructured in a nonminimal state space to achieve output feedback without a state estimator. The proposed output feedback adaptive controller is fully consistent with its state-feedback counterpart. Simulation results for tracking different reference signals demonstrate the efficacy of the proposed strategy.

This article investigates the autonomous exploration problem of an unmanned aerial vehicle (UAV) in a fully unknown three-dimensional (3D) space, subject to the constraints of collision avoidance, energy-saving, and computation consumption. To tackle this problem, a hybrid planning algorithm named FSHP is proposed. The algorithm consists of a novel local planner designed to explore unknown space within the onboard camera’s field of view (FoV) faster and less computationally. The local planner is a combination of the frontier-based and sampling-based methods, overcoming the bottlenecks of high computational time for the former and non-heuristics for the latter. Furthermore, the algorithm incorporates a global planner based on historical information to enhance performance in larger and more complex scenarios. The global planner includes a historical road map (HRM) using the rapidly-exploring random tree (RRT) and a historical tree (HST) based on the k-dimension (k-d) tree, built simultaneously. When no informative viewpoints are nearby, the planner replans trajectories globally to unexplored space. Finally, the proposed approach is evaluated in both simulations and real-world experiments, demonstrating the effectiveness and efficiency of the FSHP.Note to Practitioners—The motivation of this paper stems from the need to develop a fast and efficient autonomous exploration algorithm for a UAV for practical applications such as 3D reconstruction, search-and-rescue and military reconnaissance. Frontier-based and sampling-based methods are widely used to solve this problem due to their heuristics and low computational effort, respectively. However, either method can not meet the requirements related to exploration efficiency arising from increasingly complex and diverse tasks. To speed up the exploration process, reduce the exploration time and shorten the exploration path length, we propose this new method FSHP. It combines the advantages of global exploration (frontier-based methods) and local exploration (sampling-based methods) with random sampling in the frontiers. Furthermore, the replanning target selection and waypoints optimization schemes helps in reducing the path. Overall, this novel framework, FSHP, enables efficient and effective autonomous exploration tasks.

Urban rail transit (URT) is an essential part of urban public transportation. Accurate ridership prediction is increasingly important for the safe operation and efficient management of URT. However, existing studies regard the URT stations with different intersecting subway lines as a whole, which ignores the internal spatial connections within the stations. In fact, URT stations are embodiments of spatial coupling between subway lines. Additionally, the intrinsic patterns of ridership are also neglected. To further improve the prediction accuracy, this study proposes a deep learning model based on graph convolutional network (GCN) and bidirectional long short-term memory network (Bi-LSTM) with a non-parallel structure (D-BLGCN). At the beginning, this study decouples the URT stations according to the intersecting subway lines. On the basis of spatial decoupling, different patterns of ridership are diverged into tributaries. Then, a non-parallel structure in the proposed model is designed to capture the intrinsic spatio-temporal correlations of ridership. To the best of our knowledge, this is the first time that the integration of internal spatial connections and ridership diverging is employed for URT ridership prediction. Extensive experiments are conducted on Beijing URT ridership data with different time granularities. The results demonstrate that the proposed model achieves better prediction performance compared with baselines.

Traffic sign detection and recognition (TSDR) has attracted extensive studies recently due to its broad application prospect in Intelligent Transport Systems. TSDR is still challenging due to the small size of traffic signs in the image. Besides, the traffic signs in the real world exhibit a long-tailed distribution (i.e., data for most categories are scarce while for others are abundant.), which will lead to a significant performance drop of the detection framework. In this paper, we propose a novel traffic sign detection framework to address these challenging problems. In order to detect small traffic signs, we propose an effective adaptive and attentive spatial feature fusion module which learns the spatial attention map to fuse different feature maps at each scale while emphasizing or suppressing the features at different regions. This module can significantly alleviate the inconsistency among features and enhance feature representations of small objects. Furthermore, to address the long-tailed data problem, a hierarchical group softmax head which constructs a label tree to divide categories into different groups is proposed, in this way, categories in each group have relatively similar frequencies, then the softmax is applied in each relatively balanced group to calculate the probability of each category. Extensive experiments conducted on the TT100K and GTSDB datasets demonstrate that the proposed method achieves notable improvement in both the small traffic signs and long-tailed detection problems in TSDR.

High-resolution micro traffic data are important to traffic safety and efficiency analysis. In this study, a roadside LiDAR sensor is used to collect 3D point clouds of surrounding objects. An automatic background construction and object detection method is proposed on the basis of the operation principle of the LiDAR sensor. In the algorithm, the discrete horizontal and vertical angular values can be considered as coordinates of pixels in digital images, and the farthest and mean distance of each azimuth are used to construct the background dataset. Then, vehicle and pedestrian points are extracted on the basis of the distance difference of each point with the same angular value between the target frame dataset and the background dataset. A density-based spatial clustering method is employed to group points into clusters and identify vehicles and pedestrians automatically. Finally, the performance of the algorithm is tested for roadside LiDAR data preprocessing under different traffic conditions. Our algorithm can perform well with high accuracy. Experimental results demonstrate that our algorithm can achieve a larger detection range and exhibit lower time complexity in comparison with a previously proposed algorithm.

Intelligent Transportation Systems (ITS) are widely researched to improve the traffic situation. In ITS, vehicle detection system plays a significant role. At present, vehicle detection is often conducted by inductive loops, which are very expensive and inconvenient to install and maintain. Video camera is another frequently used detector, but it needs high computing power. In order to solve these problems, this paper focuses on the development of a roadside magnetic sensor system for vehicle detection. The device is installed at the side of the road and measures traffic in the adjacent lane (the closest lane to the sensor node). The data are transmitted by the IEEE 802.15.4 communication protocol. A novel adapted threshold state machine algorithm is proposed to detect vehicles. Since false judgments created by large vehicles passing in the nonadjacent lane (the lane next to the closest lane to the sensor node) are frequent in urban environments, a novel feature extracted by fusion of three magnetic sensor signals are proposed to reduce this error. The developed magnetic sensor system is wireless, compact, and cost-effective. The experimental results show that the proposed system achieves high accuracy and is viable in urban environments.

ABSTRACT In social groups, complex group behavior often emerges from the local interaction among simple individuals. Throughout this study, we assume that individuals can access information by way of wireless communication. In this case, individuals are able to exchange accurate information with each other as long as wireless communication links between them are allowed. Subsequently, we propose a consensus decision‐making model for studying the consistency of group behavior considering both the wireless communication range and the probability of successful communication. Our simulation results show the following conclusions: (i) consistency of the group behavior is absolutely achieved, when the wireless communication range is large enough and the probability of successful communicationp = 1; (ii) when the wireless communication range is large enough, the consistency of the group behavior is still achieved as long aspis bigger than some small constant; and (iii) the law in (iii) remains applicable when the number of individuals in the group changes. Therefore, one may infer that consistency of group behavior in mobile sensors is much more related to the extent of distribution of obtainable information than the amount of information, where the extent of distribution of obtainable information means that each individual can obtain information from wider area or from those individuals who are not only in its local area. Copyright © 2012 John Wiley & Sons, Ltd.

ABSTRACT In this study, a deterministic deployment problem in wireless sensor networks is examined. On the basis of information coverage, we study equilateral triangle and square deployment strategies, and we provide the maximum distance between sensors in order to reach the required detection probability for any point in the monitoring field. First, we provide a model of the signal attenuation. On the basis of the detected signal from the K sensors, the best linear and unbiased estimation is used to estimate the signal parameter with the corresponding error. For the equilateral triangle deployment, the maximum distance between sensors is computed and provided when the received signal data from two or three sensors is used. Similarly, we have computed and supplied the maximum distance between sensors in the square deployment. Simulations are performed to show the relationship between the number of sensors and the detection probability. The simulation results show that it is not a good choice to improve the detection probability with a larger number of sensors.Copyright © 2012 John Wiley & Sons, Ltd.