Abstract:To address the trade-off between solution efficiency and trajectory feasibility in traditional Radau Pseudo-spectral Method (RPM) for trajectory optimization, this paper proposes an adaptive threshold-based segmented polynomial-degree elevation pseudo-spectral method (AT-RPM). This approach aims to enhance nonlinear optimization accuracy and accelerate convergence. By dynamically comparing the maximum value of the error matrix with a standard time step during iterative process, the method establishes new segment points in intervals where the error exceeds an adaptive threshold. Furthermore, the number of collocation points within segments is also increased when the error falls below a deviation threshold. This strategy selectively optimizes both smooth and non-smooth regions of the trajectory solution, thereby enabling achievement of trajectory planning goals with fewer segments and collocation points. Compared with the traditional Radau Pseudo-spectral Method, AT-RPM can converge to the desired result in fewer iteration rounds, effectively balancing numerical accuracy and solution efficiency. To validate performance, comparative simulation experiments have been conducted in a non-cooperative target capture scenario involving a free-flying space robot. The results indicate that the proposed method outperforms the existing approaches in both computation time and terminal accuracy, demonstrating superior overall performance.

Abstract:To address the problem of inefficient collaborative area search by heterogeneous unmanned systems in underground spaces, this paper comprehensively considers both aerial and ground obstacles and constructs a three-dimensional grid model of the underground space. Based on this, using a three-dimensional sensor model for unmanned systems with adaptive height, the impact of detection distance on detection performance is quantitatively analyzed. Furthermore, by using the pheromone map mechanism, the environmental information is dynamically updated through the diffusion and evaporation of pheromones. Within the framework of distributed model predictive control (DMPC), an improved artificial lemming algorithm-pheromone map (IDALA-PM) is proposed, which integrates differential variation, triangular walk, Gaussian perturbation, and t-distribution adaptive perturbation strategies, to achieve distributed real-time path planning for heterogeneous unmanned aerial and ground systems. Simulation results indicate that the proposed IDALA-PM algorithm can effectively accomplish underground space search tasks and improves search efficiency by more than 54.2% compared to traditional algorithms.

Abstract:This paper investigates the intrinsic connection between neural networks, reinforcement learning (RL) algorithms, and the evolutionary principles of higher animals by developing an observable and interpretable autonomous control system for a wheel-legged robot. Leveraging the Deep Deterministic Policy Gradient (DDPG) algorithm, an Actor-Critic neural network has been implemented directly on Field-programmable gate arrays (FPGA). An FPGA-ARM robot control system is further designed to export weight activation signals in real time and generate weight heatmaps, thereby visualizing the strategy evolution process. Experimental results demonstrate that the proposed system has the ability of reducing the single-step computation latency to 28 μs and achieves convergence within 5 000 steps. Moreover, the weight heatmaps reveal the dynamic evolution of strategies across three phases——early, middle, and late stages. Qualitative analysis indicates that non-salient regions have minimal impact on the overall strategy, resulting in more efficient resource utilization. The proposed hardware-algorithm co-design framework establishes a novel paradigm for improving the interpretability and reducing the “black-box” nature of RL. It also showcases the unique advantages of FPGA in embedded robot control, namely low latency, high parallelism, and low power consumption. This work lays a robust foundation and presents promising prospects for real-time skill learning and hardware acceleration in scenarios involving multi-agent cooperation and heterogeneous computing platforms.

Abstract:To address the autonomous evasion problem for variable-sweep wing aircraft in dynamic intercept environments, this paper proposes an intelligent morphing decision algorithm. This algorithm leverages dynamic morphing, primarily through real-time adjustment of the sweep angle, as the core evasion strategy.Initially, aerodynamic coefficients for the variable-sweep-angle aircraft model are fitted using the least-squares method. The influence of these aerodynamic parameters on the aircraft′s performance is then analyzed, providing the foundation for intelligent morphing decision-making. Subsequently, a dynamic game scenario is developed for a penetration mission involving the morphing aircraft and dual interceptors, incorporating practical physical constraints such as flight speed and operational area boundaries. A decision model is then designed, integrating a state space that includes aircraft status, interceptor status, and target information, with the optimization objectives of maximizing evasion effectiveness and aerodynamic performance.Finally, simulation results demonstrate that the proposed algorithm successfully achieves autonomous morphing-based evasion while maintaining high maneuverability and agility. This approach overcomes the limitations of traditional morphing strategies, which relay on offline optimization and predefined task switching, making it difficult to adapt to highly dynamic game environments.

Abstract:Self-reconfigurable wave-like crawling (SWC) robots, characterized by their unique serial/parallel connection states, imposes stringent requirements for generating continuous and feasible trajectories during motion planning. Conventional motion planning algorithms suffer from inefficiency and fail to satisfy kinematic constraints. To address these limitations, this paper presents an enhanced motion planning method termed as informed optimal rapid-exploration random tree. First, the rapidly-exploring random tree connect (RRT-Connect) algorithm is employed to generate an initial feasible path and construct an elliptical state-space sampling domain, facilitating rapid expansion of the random tree. Besides, polynomial trajectories are optimized based on a minimized jerk objective function and Hessian matrix to generate smooth motion profiles that conform to the kinematic characteristics of the SWC robot. Finally, simulations conducted in diverse obstacle scenarios validate the effectiveness of the optimized algorithm. The results demonstrate that, compared to traditional algorithms, the proposed method significantly enhances path planning efficiency across various obstacle environments, reducing global sampling time and shortening planned path length. Furthermore, it effectively mitigates abrupt acceleration and sharp turns during the robot’s movement, yielding paths that are better suited for practical operational applications.

Abstract:To enhance the efficiency and flexibility of logistics distribution under spatiotemporal motion coupling constraints, a heterogeneous aerial-ground collaborative intelligent transportation system is developed, which integrates the high mobility of unmanned aerial vehicles with the strong carrying capacity of unmanned ground vehicles. A multi-layer multi-scale multi-robot transportation trajectory planning (M³TTP) method is designed, which combines an improved ant colony optimization (ACO) algorithm with the probabilistic roadmap (PRM) algorithm. This method optimizes the trajectory solution for the traveling salesman problem (TSP) under multi-objective obstacle constraints, to achieve multi-scale hierarchical task allocation and decision planning.The simulation results demonstrate that, considering the real-world spatial obstacle and dynamic time sequence constraints, an adjacency matrix is constructed using an improved terminal heuristic algorithm. This enables the planning of an optimal transportation path based on detailed information from a small-scale and well-known road network.A time synchronization mechanism is proposed to address the challenges of heterogeneous collaborative path planning. The discrete Dubins path planning algorithm is presented to improve the efficiency of collaborative tasks. The proposed M³TTP method enhances both the adaptability and delivery efficiency of the system under spatiotemporal motion coupling constraints.

Abstract:To address the constraint conflict issues in the prescribed performance tracking control of nonlinear systems and simultaneously impose more refined constraints on the system, this paper proposes a constrainted performance control method based on a novel variable barrier function. Firstly, the problem of constraint conflicts caused by uncertain initial states and sudden changes in the expected trajectory in the constraint control of nonlinear systems is analyzed, and a new asymmetric variable barrier function configuration is proposed. This configuration introduces an attraction function to the system, which can uniformly handle the two stages of constraint release and constraint recovery during the constraint conflict process, and enrich the setting form of the preset performance boundary, making the constraint control strategy simpler and more complete. Secondly, based on the designed novel asymmetric variable barrier function configuration, a set of new L-shaped preset performance boundary functions is designed. These performance boundary functions are inherently constraining, capable of imposing quantitative constraints on the overshoot of the system output, thereby more precisely confining the system behavior. Finally, the boundedness of the system error and the forward invariance are proved, and the uniform ultimate boundedness of all closed-loop signals is proved based on the Lyapunov stability theory. A quadrotor unmanned aerial vehicle system is selected for numerical simulation and experimental comparison. The final numerical simulation and experimental results verify the effectiveness and superiority of the proposed scheme. The designed attraction-preserving performance method in this paper can not only impose quantitative overshoot constraints on the system, but also solve the constraint conflict problems arising during the constraint process.

Abstract:Hyperparameter optimization (HPO) is a pivotal technology in Automated Machine Learning, aiming to automate the tuning process and alleviate the burden on practitioners. In robotic systems, HPO plays a critical role in enhancing neural network training for perception modules, controller parameter calibration, and performance optimization of multimodal data fusion algorithms. However, despite significant progress, efficiency remains the primary bottleneck limiting its widespread adoption. Recent advances in meta-learning have opened new avenues for improving HPO efficiency, particularly demonstrating unique advantages in robotic systems that require rapid adaptation to dynamic environments and novel task scenarios. This technique enables models to automatically assimilate and apply knowledge from prior tasks, thereby improving learning efficiency for unseen tasks. Currently, researchers are actively exploring meta-learning techniques to enhance HPO search capabilities. In this paper we aim to provide a systematic overview of relevant research. First, we provide a formal definition of the HPO problem and review state-of-the-art methods. Subsequently, we systematically summarize meta-learning-based HPO strategies and analyze prevailing meta-learning algorithms. Furthermore, we introduce benchmark datasets in HPO research and compare the performance of mainstream methods. Finally, we discuss future research directions in hyperparameter optimization technology.

Abstract:To address the issues of task offloading instability, underutilization of computing resources, high offloading costs, and low user service quality in vehicle edge computing, which are caused by highly dynamic network topologies, high-dimensional action exploration spaces, and strict low-latency constraints, a digital twin-assisted clustering method for vehicle edge computing task offloading is proposed. Firstly, a vehicle social relationship model is constructed to quantify the relationships between vehicles. By introducing two social trust factors to measure offloading stability between vehicles, the method reduces the waste of computing resources caused by unstable communication links. Next, a digital twin-vehicle edge computing network model is constructed. Through the bidirectional information interaction between the digital twin network model and the physical vehicle edge device network, the condition of vehicle edge devices is monitored in real time. In addition, a clustering algorithm based on the gravity model is designed to assist vehicle clustering, thus narrowing the space for action exploration, improving the efficiency of computing task offloading, and reducing the cost of edge task computation. Finally, based on the above optimization strategies, a dual-latency deep deterministic policy gradient edge computing task offloading algorithm assisted by digital twin clustering is designed and implemented. Simulation experiments demonstrate that, compared to existing offloading methods, the proposed method significantly reduces task offloading costs, decreases task execution latency, and improves task offloading success rates while fully utilizing computing resources. This enables the efficient and stable offloading of vehicle-edge computing tasks, thereby enhancing user service quality.

Abstract:To address key limitations in existing community search methods for heterogeneous information networks (HINs) , this paper proposes a community search method for HINs that integrates multiple semantic relationships, employing an efficient “offline learning, online search” strategy. Its core lies in: adaptively learning the weight contributions of different meta-paths to the target community cohesiveness using a semantic attention mechanism to precisely quantify semantic differences; and subsequently measuring node relevance by combining network structure and node attribute features to locate community members. In the offline phase, a node-community association model is pre-trained to generate probability distribution vectors indicating node affiliation across various communities. In the online phase, community search is rapidly responded to based on precomputed results. This strategy maintains the flexibility of the learning model to effectively capture heterogeneous network semantics and attributes, while shifting the main computational burden to the offline phase, significantly improving query efficiency, making it particularly suitable for high-query-frequency scenarios. Experiments on multiple real-world HIN datasets demonstrate that our method significantly outperforms existing mainstream methods in both community effectiveness (semantic relevance, structural cohesiveness, attribute consistency) and query efficiency.

Abstract:To improve the accuracy of image time-series prediction, a time-series prediction network of MA-LSTM is proposed based on LSTM and attention mechanism. This model is consist of multi-scale attention module (MAB), multi-scale attention layer (MALayer) and super-resolution reconstruction module (SRRM), it could improve the express spatiotemporal features and long-range dependencies. Firstly, MAB module is designed, and detail modeling is improved through the spatiotemporal feature enhancement layer (GSTA), then the channel feature enhancement layer (GCA), overcoming SwinLSTM′s insufficient capture of fine-grained features, is used to enhance the cross-dimensional information interactions of the feature map. Secondly, a simplified LSTM structure is employed, and MALayer is constructed in combination with MAB to improve modeling of time series information. Finally, the SRRM module is designed during feature map reconstruction to improve the prediction output. Experimental results show that MA-LSTM achieves a structural similarity index(SSIM) of 0.960 2 and 0.924 3 on two datasets in different fields: MovingMNIST and KTH. Compared with SwinLSTM, PhyDNET, PredRNN, and ConvLSTM networks, the highest accuracy improvement of 0.337 and 0.212, respectively. This model demonstrates the higher efficiency and applicability in time series prediction tasks and the well potential for cross-domain promotion, and the ablation experiments also show the effectiveness of the proposed module.

Abstract:To improve the convergence speed and robustness of in-phase quadrature (IQ) imbalance correction in zero-IF receivers, this study integrates the Kalman filtering algorithm with a blind source separation structure and proposes a dual-channel Kalman filter-based correction method. By leveraging state-space modeling and adaptive covariance updates, the proposed algorithm enables more efficient and stable parameter estimation in dynamic environments, thereby achieving effective compensation for IQ mismatch. Comparative simulations were conducted between the proposed algorithm and the least mean square (LMS), normalized least mean square (NLMS), and affine projection algorithm (APA). The results show that the image rejection ratio (IRR) of the corrected signals reaches approximately 45 dB for all methods. However, the IRR surface of the proposed dual-channel Kalman filtering algorithm is smoother. Additionally, under 16QAM and 16PSK modulation schemes, the proposed algorithm yields the lowest symbol error rate (SER), indicating improved correction performance and stability. The algorithm also demonstrates superior convergence, with its mean squared error (MSE) approaching zero after approximately 50 iterations, while LMS, NLMS, and APA require about 0,0, and 200 iterations, respectively, to converge. Furthermore, sensitivity analysis reveals that the IRR variation remains minimal across a wide range of parameter settings, demonstrating the robustness of the proposed algorithm.

Abstract:To accurately evaluate the bearing and deformation characteristics of a four-pile jacket foundation under complex environmental loads, a three-dimensional numerical model was developed based on the improved nonlinear hysteretic p(soil resistance) -y (lateral displacement of the pile) curve, with P-multipliers incorporated to account for pile-group effects and axial forces. Furthermore, a two-point combined cyclic loading approach was adopted to represent wind and wave loads, respectively, taking into consideration differences in load amplitude, frequency, and application height. Results indicate that when the pile spacing is ≥7D, the group pile effect can be negligible, and as the loading height increases (constant loading displacement), the influence of axial forces on the foundation’s lateral behaviors is strengthened. An increase in the wave load ratio and a decrease in the wind load height significantly amplify the deformation and bending moment of the foundation, and it is more sensitive to variations in the load ratio. Under small-amplitude cyclic loading, the pile head displacement increases linearly with the cycle number (in logarithmic coordinates), and the cumulative displacement tends to stabilize. However, at a loading amplitude of 0.4, the plastic cumulative displacement continues to increase.

Abstract:To address the limitation that existing seismic isolation devices mainly focus on horizontal directions while providing insufficient vertical control, a novel hydraulic accumulator-based vertical seismic isolation system was proposed to mitigate the vertical seismic response of the structures. The system′s stiffness characteristics were analyzed, followed by experimental validation to assess its mechanical performance and energy dissipation capability. A dynamic model of the isolation system was then established, and the harmonic balance method was employed to determine the system′s amplitude-frequency response and displacement transmissibility. The effects of vibration amplitude, damping ratio, and accumulator volumetric compression coefficient on the displacement transmissibility were analyzed. Finally, finite element models for both the vertical seismic isolation structure and a conventional frame were established and subjected to seismic time history analysis, and the seismic response results of the two frameworks were compared. The analysis results show that the vertical acceleration response of VSI-HCA structures is reduced by an average of 59% compared to the ground input, indicating a significant vertical isolation effect.

Abstract:This study aims to clarify the working mechanism of concrete-filled circular aluminum alloy tubular columns andverify/improve the reliability of the bearing capacity calculation models. The available experimental tests for concrete-filled circular aluminum alloy tubular stub columns under axial compression were collected and analyzed in this paper, with the experimental database established. The mechanical properties of the concrete-filled circular aluminum alloy tubular stub columns under axial compression were further investigated, and the applicability of the existing five typical bearing capacity calculation models was evaluated. A new calculation model with higher accuracy for predicting the compressive bearing capacity of concrete-filled circular aluminum alloy tubular stub columns was proposed. The investigated results indicate that some of the test variables of the existing studies may be beyond the scope of “steel-concrete composite structures”, which should be analyzed individually. The boundary confinement factor of concrete-filled circular aluminum alloy tubular columns is higher than that of conventional concrete-filled circular steel tubular columns under the same condition which can be taken as 1.75, approximately. The five typical existing calculation models can be used to provide an approximate prediction of the compressive bearing capacity of concrete-filled circular aluminum alloy tubular stub columns. However, there are still 5.6% to 32.4% of the collected specimens with predicted deviations exceeding 20%, which are mainly concentrated in the range of lower confinement factors (ξ<1.0). The proportion of the collected test specimens with deviations beyond 20% in the compressive bearing capacity predicted by the proposed calculation model in this paper is less than 3%.

Abstract:To reduce the residual deformation of structures under seismic loading, a self-centering rotary friction damper (SMA-SRFD) based on shape memory alloy (SMA) bolts was proposed. The basic structure and working principle of the damper were introduced. Material property tests of SMA bolts were conducted, and theoretical analysis of the force performance of the bolts and SMA-SRFD was carried out. The results were compared with those from finite element analysis to verify the accuracy of the theoretical formulas used. A three-dimensional finite element model of the SMA-SRFD was established, and a parametric analysis was performed with the rotation angle of the frictional inclined plane, the friction coefficient, and the pre-tension of the SMA bolts as variables. The results show that the stress-strain curve of the SMA bolt exhibits a typical “flag” shape, indicating good energy dissipation and self-centering capabilities. The results of theoretical analysis and numerical simulation of the SMA-SRFD are consistent, both accurately reflecting the force process of the damper. Increasing the inclination angle of the inclined plane enhances the load-bearing capacity and equivalent stiffness of the SMA-SRFD, but the energy dissipation and damping ratio decrease. By increasing the coefficient of friction between the contact surfaces, the energy dissipation, equivalent stiffness, and equivalent damping ratio of the SMA-SRFD all increase, while the self-centering capability weakens. As the pre-tension of the SMA bolts increases, the maximum load-bearing capacity of the SMA-SRFD remains essentially the same, while the deformation decreases slightly. Moreover, the different pre-tensions to the SMA bolts does not significantly change the hysteresis performance of the damper.

Abstract:To study the bond-slip performance between ribbed steel bars and grout under Type II APC sleeve restraint, a total of 69 Type II APC sleeve-constrained grout reinforcement tensile tests in 23 groups were conducted in this study. An electro-hydraulic servo universal testing machine was used for unidirectional tensile loading. Strain gauges and displacement meters were used to collect data on the strain of the sleeve and the slip value of the reinforcing bar, respectively. The damage morphology of the specimens, factors affecting bond strength, bond-slip ontological relationship, and sleeve load-strain curve were studied. The results show that: The damage forms of the sleeve-constrained grout specimens include two types: the pull-out damage of steel bars (before and after yielding) and the pull-out damage of steel bars; The bond strength of the specimens decreases with the increase of the diameter of the steel bars and the anchorage length, and increases with the increase of the steel content rate; The bond slippage between the steel bars and the constrained grout is divided into four phases: slippage, cleavage, descending, and remnant;Comparison of anchorage specimens made of different materials reveals that when reaching the ultimate bond strength the slip values decrease in the order of unrestrained grout, sleeve-constrained grout, and concrete, and the slip value of concrete was the smallest due to the limitation of crack development by the aggregates. Energy analysis of the bond-slip curves shows that the ductility of the specimens is superior, and the coefficient of brittleness decreases with increasing reinforcement diameter and increases with increasing anchorage length. At the ultimate load, the annular direction of the long side of the sleeve is tensile strain, and it increases with the increase of the diameter of the reinforcement bar, the annular direction of the short side is mostly compressive strain, the longitudinal strain of the long side and short side of the sleeve is tensile strain, and the longitudinal and annular strain of the long side of the sleeve is larger than that of the short side. Based on the ABAQUS platform, a finite element model of type II APC sleeve-constrained grout was constructed, and the simulation results were in good agreement with the test data. When calculating the value of critical anchorage length of reinforcement in the critical state of yielding and pulling off, the existence of the sleeve significantly reduces this length.