In dynamic systems, complex time-varying noise characteristics impose challenges for multi-sensor data fusion. This paper proposes a deep learning-driven noise-adaptive filtering approach for multi-sensor data fusion. A parallel gated recurrent unit (GRU) and one-dimensional convolutional neural network (1D-CNN) architecture jointly extracts temporal dependencies and local spatial patterns from raw measurement sequences to estimate sensor noise variances. Each sensor then uses these variance estimates for more accurate adaptive local filtering. Based on these refined local estimates, the fusion center employs an event-triggered scheme to drastically cut communications while preserving fusion accuracy. Comprehensive experiments in a high-fidelity Unreal Engine–AirSim quadrotor unmanned aerial vehicle (UAV) simulation under nominal, drift, abrupt, and extreme noise modes confirm that our pipeline achieves superior global state estimation while dramatically reducing data transmission burden.

Ground-to-air communication is a critical technology for establishing an Internet of Things (IoT) network system, especially in emergency situations. We are investigating the trajectory planning problem of a data collection IoT network assisted by an unmanned aerial vehicle (UAV). This article aims to solve the data collection Dubins traveling salesman problem (DCDTSP) for UAVs in a three-dimensional and complex obstacle environment. To optimize the paths for UAVs in data collection from terminals to UAVs, a novel releasing-collecting-recycling (RCR) framework has been established for heterogeneous multi-UAVs. In the UAV release step, we propose a multi-height hierarchical target clustering (MHTC) algorithm to enhance the efficiency of multi-target clustering. In the data collection step, a bundling ant colony system (BACS) is developed to minimize the length of the obstacle avoidance path while still meeting the communication throughput constraint. Meanwhile, the dynamic adaptive window probabilistic roadmap (DAWPRM) algorithm has been enhanced to address the obstacle avoidance distance in BACS. In the UAV recycling step, we propose a time synchronous Dubins recycling strategy to plan the simultaneous arrival trajectory for multiple UAVs with a constrained turning radius. The results of simulation experiments showed that the proposed RCR framework is optimal for finding Pareto solutions for DCDTSP.

Jianguo Zhang, Thai Quoc Hoang, Ming Zhu, Zuxin Liu, Shiyu Wang, Tulika Manoj Awalgaonkar, Akshara Prabhakar, Haolin Chen, Weiran Yao, Zhiwei Liu, Juntao Tan, Juan Carlos Niebles, Shelby Heinecke, Huan Wang, Silvio Savarese, Caiming Xiong. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 2025.

Large language model (LLM) agents have shown great potential in solving real-world software engineering (SWE) problems. The most advanced open-source SWE agent can resolve over 27% of real GitHub issues in SWE-Bench Lite. However, these sophisticated agent frameworks exhibit varying strengths, excelling in certain tasks while underperforming in others. To fully harness the diversity of these agents, we propose DEI (Diversity Empowered Intelligence), a framework that leverages their unique expertise. DEI functions as a meta-module atop existing SWE agent frameworks, managing agent collectives for enhanced problem-solving. Experimental results show that a DEI-guided committee of agents is able to surpass the best individual agent's performance by a large margin. For instance, a group of open-source SWE agents, with a maximum individual resolve rate of 27.3% on SWE-Bench Lite, can achieve a 34.3% resolve rate with DEI, making a 25% improvement and beating most closed-source solutions. Our best-performing group excels with a 55% resolve rate, securing the highest ranking on SWE-Bench Lite. Our findings contribute to the growing body of research on collaborative AI systems and their potential to solve complex software engineering challenges.

Jianguo Zhang, Tian Lan, Ming Zhu, Zuxin Liu, Thai Quoc Hoang, Shirley Kokane, Weiran Yao, Juntao Tan, Akshara Prabhakar, Haolin Chen, Zhiwei Liu, Yihao Feng, Tulika Manoj Awalgaonkar, Rithesh R N, Zeyuan Chen, Ran Xu, Juan Carlos Niebles, Shelby Heinecke, Huan Wang, Silvio Savarese, Caiming Xiong. Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2025.

Training effective AI agents for multi-turn interactions requires high-quality data that captures realistic human-agent dynamics, yet such data is scarce and expensive to collect manually. We introduce APIGen-MT, a two-phase framework that generates verifiable and diverse multi-turn agent data. In the first phase, our agentic pipeline produces detailed task blueprints with ground-truth actions, leveraging a committee of LLM reviewers and iterative feedback loops. These blueprints are then transformed into complete interaction trajectories through simulated human-agent interplay. We train a family of models---the xLAM-2-fc-r series with sizes ranging from 1B to 70B parameters. Our models outperform frontier models such as GPT-4o and Claude 3.5 on $\tau$-bench and BFCL benchmarks, with the smaller models surpassing their larger counterparts, particularly in multi-turn settings, while maintaining superior consistency across multiple trials. Comprehensive experiments demonstrate that our verified blueprint-to-details approach yields high-quality training data, enabling the development of more reliable, efficient, and capable agents. We open-source both the synthetic data collected and the trained xLAM-2-fc-r models to advance research in AI agents.Dataset: https://huggingface.co/datasets/Salesforce/APIGen-MT-5k & Models: https://huggingface.co/collections/Salesforce/xlam-2-67ef5be12949d8dcdae354c4

This paper studies a fractional-order control and stability scheme for a class of n-DOF mechanical systems that are subject to actuator saturation and unknown disturbances. By the fractional-order Leibniz rule, a generalized fractional-order Lyapunov stability (FOLS) method is developed to support the design of a class of fractional-order robust controllers, which can guarantee the uniform ultimate boundedness and faster convergence. Based on this, a fractional-order backstepping-like feedback controller (FOBFC) is raised to realize an effective and stable control performance of mechanical systems, which can ameliorate the shortages of the traditional backstepping control effectively. Moreover, a set of fractional-order methods, including actuator compensation (FOAC) and disturbance observer (FODO), are proposed and involved in FOBFC to handle the actuator saturation and unknown time-varying disturbances, making it possible to achieve a more robust and stable control performance. Finally, numerical simulations and comparative experiments are carried out to demonstrate the effectiveness and superiority of the proposed fractional-order control scheme, in n-DOF mechanical systems. Note to Practitioners—A fractional-order control and stability scheme is developed to regulate a class of practical n-DOF mechanical systems, which is composed of FOBFC, FOAC and FODO. It can be applied to deal with the actuator saturation and time-varying disturbance of n-DOF mechanical systems, contributing to realizing a more robust and stable control performance. In the application, a Euler-Lagrange dynamics of mechanical systems is required for constructing the fractional-order tracking errors. Subsequently, FOBFC law can be raised based on the generalized FOLS, to guarantee the asymptotic convergence and stability of mechanical systems. To address the practical issues of mechanical systems, FOAC and FODO are necessary for the proposed control scheme to compensate for the actuator saturation adaptively, and handle the unknown and time-varying disturbances effectively.

Dynamics-based control offers a promising approach to exploring the motion potential of soft robots. However, inherently infinite degrees of freedom of these systems pose significant challenges for dynamics modeling, closely followed by the pressing robustness concerns arising from finite-dimensional approximations. This paper addresses these issues by proposing a physics-informed dynamics learning neural network and an adaptive fractional-order control for continuum soft robots. Specifically, a deep Lagrangian neural network is first developed with an embedded self-attention mechanism to enhance learning efficiency, accuracy, and data sensitivity. Subsequently, an adaptive fractional-order sliding mode controller is designed, leveraging the inherent historical memory properties of fractional calculus. This controller not only ensures robust shape control but also improves response speed and tracking accuracy. To further handle model discrepancies in the learned dynamics and external disturbances, a nonlinear disturbance observer is introduced to effectively estimate and compensate for lumped uncertainties, thereby ensuring reliable performance. Theoretical analysis confirms the closed-loop stability, while both simulation and experiment results validate the high dynamics fitting accuracy of the proposed network, as well as the robust and precise tracking capability of the fractional-order controller. Note to Practitioners—Soft robots offer great potential in unstructured or constrained environments owing to their compliance and adaptability. However, their high degrees of freedom and nonlinear behaviors make analytical modeling and robust control particularly challenging. Meanwhile, traditional closed-box learning methods often suffer from limited physical interpretability, reliability and extrapolability. This work presents a physics-informed dynamics learning framework combined with a fractional-order controller for soft robots. The dynamics learning network embeds physical priors to enhance model interpretability and extrapolability, while a self-attention mechanism improves data efficiency and modeling accuracy. Additionally, a disturbance observer is designed to estimate and compensate for model discrepancies and external disturbances, thereby contributing to the system’s robustness. Incorporating the observer’s outputs, the adaptive fractional-order controller further enhances closed-loop behavior by leveraging the memory properties of fractional calculus.

This paper is focused on the event-based secure control issue for cyber-physical systems (CPSs) under aperiodic denial-of-service (DoS) attacks. Malicious DoS attacks disrupt the communication between the controller and the actuator. The finite attack resources of malicious attackers are taken into consideration, and the DoS attacks are characterized using an aperiodic model. In contrast to prior results, the present study tackles the issue of secure controller design by considering the attributes of the DoS attack, instead of employing a switched system approach to address the aforementioned concerns. More specifically, under aperiodic DoS attacks, sufficient criteria are established to guarantee that the closed-loop CPSs can achieve bounded stability. Then, within a time-varying attack period, the relationship between the attack active interval and the attack silent interval is derived. Without satisfying the derived conditions, the system’s stability will deteriorate. Moreover, an event-based secure control scheme under aperiodic DoS attacks is designed. To verify the efficacy of the derived theory, a wheeled mobile robot system under aperiodic DoS attacks is illustrated. Note to Practitioners—CPSs have been widely utilized in various domains, such as aerospace and intelligent transportation. However, the openness of networks provides attackers with numerous opportunities for malicious assaults, consequently leading to a degradation in system performance. Consequently, researching the security issues of CPSs under malicious attacks is of utmost urgency. This paper focuses on the issue of event-triggered secure control for CPSs in the presence of energy-constrained aperiodic DoS attacks. The event-triggered communication mechanism is introduced to reduce the computational burden. The criteria for ensuring the bounded stability of CPSs under aperiodic DoS attacks are proposed. The relationship between the attack active interval and the attack silent interval is derived, which is incorporated into the proposed criteria. A wheeled mobile robot system is given to validate the effectiveness of the proposed method. In the future, an active defense control method will be proposed to counter malicious attacks.

Recent months have seen the emergence of a powerful new trend in which large language models (LLMs) are augmented to become autonomous language agents capable of performing objective oriented multi-step tasks on their own, rather than merely responding to queries from human users. Most existing language agents, however, are not optimized using environment-specific rewards. Although some agents enable iterative refinement through verbal feedback, they do not reason and plan in ways that are compatible with gradient-based learning from rewards. This paper introduces a principled framework for reinforcing large language agents by learning a retrospective model, which automatically tunes the language agent prompts from environment feedback through policy gradient. Specifically, our proposed agent architecture learns from rewards across multiple environments and tasks, for fine-tuning a pre-trained language model which refines the language agent prompt by summarizing the root cause of prior failed attempts and proposing action plans. Experimental results on various tasks demonstrate that the language agents improve over time and that our approach considerably outperforms baselines that do not properly leverage gradients from the environment.

Identifying the underlying time-delayed latent causal processes in sequential data is vital for grasping temporal dynamics and making downstream reasoning. While some recent methods can robustly identify these latent causal variables, they rely on strict assumptions about the invertible generation process from latent variables to observed data. However, these assumptions are often hard to satisfy in real-world applications containing information loss. For instance, the visual perception process translates a 3D space into 2D images, or the phenomenon of persistence of vision incorporates historical data into current perceptions. To address this challenge, we establish an identifiability theory that allows for the recovery of independent latent components even when they come from a nonlinear and non-invertible mix. Using this theory as a foundation, we propose a principled approach, CaRiNG, to learn the Causal Representation of Non-invertible Generative temporal data with identifiability guarantees. Specifically, we utilize temporal context to recover lost latent information and apply the conditions in our theory to guide the training process. Through experiments conducted on synthetic datasets, we validate that our CaRiNG method reliably identifies the causal process, even when the generation process is non-invertible. Moreover, we demonstrate that our approach considerably improves temporal understanding and reasoning in practical applications.

The advancement of function-calling agent models requires diverse, reliable, and high-quality datasets. This paper presents APIGen, an automated data generation pipeline designed to synthesize high-quality datasets for function-calling applications. We leverage APIGen and collect 3,673 executable APIs across 21 different categories to generate diverse function-calling datasets in a scalable and structured manner. Each data in our dataset is verified through three hierarchical stages: format checking, actual function executions, and semantic verification, improving its reliability and correctness. We demonstrate that models trained with our curated datasets, even with only 7B parameters, can achieve state-of-the-art performance on the Berkeley Function-Calling Benchmark, outperforming multiple GPT-4 models. Moreover, our 1B model achieves exceptional performance, surpassing GPT-3.5-Turbo and Claude-3 Haiku. We release a dataset containing 60,000 high-quality entries, aiming to advance the field of function-calling agent domains. The dataset and models are available on the project homepage \url{https://apigen-pipeline.github.io/}.

In multi-vehicle systems, the coupling problem between the task allocation and path planning and the variability of task execution solutions creates challenges for utility estimation and affects the effectiveness of distributed mission planning. To characterize the effect of task sequences on the task utilities and implement a task-extended distributed allocation, we propose a task-extended utility tensor algorithm (TEUTA) based on market mechanism. In the mission planning problem of multi-vehicle system, we consider the impact of the task schedule on the vehicle trajectory, and indicate the vehicle task execution utilities under different preceding task points in the form of tensors. Further, a task-extended utility tensor iterative algorithm (TEUTIA) is presented based on an iterative strategy to improve the algorithm in terms of computational complexity. A task execution utility estimation model and an algorithm framework are designed for the implementation of the two proposed algorithms. The simulation and experimental results show that compared with the non-tensor method, TEUTA and TEUTIA can achieve higher task execution performance, and TEUTIA has better computational efficiency. Note to Practitioners—This work presents two novel multi-vehicle distributed mission planning algorithms based on the market mechanism. TEUTA and TEUTIA proposed in this paper can be applied to address the impact of vehicle motion constraints on mission planning, which are widely present in various types of common nonholonomic vehicles such as two-wheel differential drive vehicles and Ackerman steering vehicles. In the application of the algorithms, an accurate kinematic model of the vehicles is required for trajectory planning to estimate the task execution reward precisely, which is necessary for effective mission planning. When the vehicle trajectory planning algorithm is computationally intensive, TEUTIA can significantly reduce the computational consumption and improve the mission planning efficiency compared with TEUTA without losing task execution reward. Finally, a stable inter-vehicle communication network is required for the interactive process of the market mechanism, where bi-directional communication exists between any two vehicles, to ensure the stability of mission planning.

This article aims to solve a hierarchical multi-UAV Dubins traveling salesman problem (HMDTSP). Optimal hierarchical coverage and multi-UAV collaboration are achieved by the proposed approaches in a 3-D complex obstacle environment. A multi-UAV multilayer projection clustering (MMPC) algorithm is presented to reduce the cumulative distance from multilayer targets to corresponding cluster centers. A straight-line flight judgment (SFJ) was developed to reduce the calculation of obstacle avoidance. An improved adaptive window probabilistic roadmap (AWPRM) algorithm is addressed to plan obstacle-avoidance paths. The AWPRM improves the feasibility of finding the optimal sequence based on the proposed SFJ compared with a traditional probabilistic roadmap. To solve the solution to TSP with obstacles constraints, the proposed sequencing-bundling-bridging (SBB) framework combines the bundling ant colony system (BACS) and homotopic AWPRM. An obstacle-avoidance optimal curved path is constructed with a turning radius constraint based on the Dubins method and followed up by solving the TSP sequence. The results of simulation experiments indicated that the proposed strategies can provide a set of feasible solutions for HMDTSPs in a complex obstacle environment.

In this article, a novel adaptive fixed-time fuzzy control algorithm is designed for uncertain Euler–Lagrange (EL) systems with actuator control input saturation. In contrast to existing algorithms, this article explores a faster fixed-time backstepping control algorithm. It enables the system to achieve fixed-time convergence with a faster convergence rate and obtain a smaller upper bound of the convergence time. To address the problem of actuator control input saturation, a novel fixed-time auxiliary system is constructed, involving coordinate transformation of the system's error variables to mitigate the effects of saturation. In response to the unknown dynamics (including model uncertainty, external disturbance, etc.) of the EL system, this article designs an adaptive interval type-2 fuzzy neural network for estimation and compensation. Stability analysis confirms that the tracking error can achieve faster fixed-time convergence. Simulation and experimental results demonstrate that the proposed control algorithm can enhance dynamic and steady-state tracking control performance.

In this article, an observer-based prescribed performance speed control method is proposed for permanent magnet synchronous motors. A transformed speed error is introduced and a suitable controller is designed to make it converge to zero, while guaranteeing the original speed error evolves strictly within a prescribed region. The controller is designed based on a backstepping approach. A linear extended state observer is applied to estimate and feed forward the external constant load disturbance to improve robustness. A data-driven radial-basis function neural network is proposed to approximate the nonlinear dynamic caused by parameter uncertainties and periodic-changing disturbance by deploying real-time and historical data. The stability analysis is based on Lyapunov's control theory. Experimental results verify the effectiveness and advantages of the proposed control scheme.

This article examines the problems of task allocation and path optimization for multiobjective coverage in a congestion risk environment with obstacle constraints. An improved probabilistic roadmap (PRM*) algorithm is proposed, which eliminates the zig-zag paths around the path endpoints. The$K$-distance PRM*$(K$-DPRM*) provides a novel clustering metric for task allocation in an obstacle environment. An ant colony system-PRM* (ACS-PRM*) algorithm is proposed to solve the congestion avoidance traveling salesman problem (CATSP) by voyage optimization of multiobjective coverage. Additionally, the mapping relationship between the probability of environmental congestion and the velocity of robot is established and combined with the feedforward control method to improve the motion control of robots. Simulations and experiments verify the effectiveness of the path optimization method in obstacle environments with congestion risk.

There exist two problems: 1) the model uncertainties caused by flexible cable and 2) the oscillations caused by the asynchronous adjustment of multiple cable lengths, which make it difficult to achieve accurate tracking control of the end-effector (EE) for cable-driven parallel robots (CDPRs) in practical applications. This paper addresses a non-singular terminal sliding mode control scheme with the relative-coupling synchronization in the cable space (NTSM-RSC) to overcome the effect of the model uncertainties and suppress the oscillations of cables simultaneously. A non-singular terminal sliding mode controller (NTSMC) is proposed to enhance the robustness of the system to model uncertainties. A relative-coupling error vector of multiple cable lengths is established based on ring topology, and a relative-coupling synchronization controller in the cable space (RSC) is proposed to improve the synchronization of multiple cable lengths. The RSC is added to the NTSMC to improve the synchronization of cables compared with the NTSMC, namely the NTSM-RSC. The finite-time convergence of the error of system is theoretically guaranteed. The effectiveness and superiority of the NTSM-RSC are verified by experiments.

In unsupervised causal representation learning for sequential data with time-delayed latent causal influences, strong identifiability results for the disentanglement of causally-related latent variables have been established in stationary settings by leveraging temporal structure. However, in nonstationary setting, existing work only partially addressed the problem by either utilizing observed auxiliary variables (e.g., class labels and/or domain indexes) as side information or assuming simplified latent causal dynamics. Both constrain the method to a limited range of scenarios. In this study, we further explored the Markov Assumption under time-delayed causally related process in nonstationary setting and showed that under mild conditions, the independent latent components can be recovered from their nonlinear mixture up to a permutation and a component-wise transformation, without the observation of auxiliary variables. We then introduce NCTRL, a principled estimation framework, to reconstruct time-delayed latent causal variables and identify their relations from measured sequential data only. Empirical evaluations demonstrated the reliable identification of time-delayed latent causal influences, with our methodology substantially outperforming existing baselines that fail to exploit the nonstationarity adequately and then, consequently, cannot distinguish distribution shifts.

The problem of learning-based control for robots has been extensively studied, whereas the security issue under malicious adversaries has not been paid much attention to. Malicious adversaries can invade intelligent devices and communication networks used in robots, causing incidents, achieving illegal objectives, and even injuring people. This article first investigates the problems of optimal false data injection attack scheduling and countermeasure design for car-like robots in the framework of deep reinforcement learning. Using a state-of-the-art deep reinforcement learning approach, an optimal false data injection attack scheme is proposed to deteriorate the tracking performance of a robot, guaranteeing the tradeoff between the attack efficiency and the limited attack energy. Then, an optimal tracking control strategy is learned to mitigate attacks and recover the tracking performance. More importantly, a theoretical stability guarantee of a robot using the learning-based secure control scheme is achieved. Both simulated and real-world experiments are conducted to show the effectiveness of the proposed schemes.

Unsupervised domain adaptation is critical to many real-world applications where label information is unavailable in the target domain. In general, without further assumptions, the joint distribution of the features and the label is not identifiable in the target domain. To address this issue, we rely on a property of minimal changes of causal mechanisms across domains to minimize unnecessary influences of domain shift. To encode this property, we first formulate the data generating process using a latent variable model with two partitioned latent subspaces: invariant components whose distributions stay the same across domains, and sparse changing components that vary across domains. We further constrain the domain shift to have a restrictive influence on the changing components. Under mild conditions, we show that the latent variables are partially identifiable, from which it follows that the joint distribution of data and labels in the target domain is also identifiable. Given the theoretical insights, we propose a practical domain adaptation framework, called iMSDA. Extensive experimental results reveal that iMSDA outperforms state-of-the-art domain adaptation algorithms on benchmark datasets, demonstrating the effectiveness of our framework.

Recently in the field of unsupervised representation learning, strong identifiability results for disentanglement of causally-related latent variables have been established by exploiting certain side information, such as class labels, in addition to independence. However, most existing work is constrained by functional form assumptions such as independent sources or further with linear transitions, and distribution assumptions such as stationary, exponential family distribution. It is unknown whether the underlying latent variables and their causal relations are identifiable if they have arbitrary, nonparametric causal influences in between. In this work, we establish the identifiability theories of nonparametric latent causal processes from their nonlinear mixtures under fixed temporal causal influences and analyze how distribution changes can further benefit the disentanglement. We propose TDRL, a principled framework to recover time-delayed latent causal variables and identify their relations from measured sequential data under stationary environments and under different distribution shifts. Specifically, the framework can factorize unknown distribution shifts into transition distribution changes under fixed and time-varying latent causal relations, and under global changes in observation. Through experiments, we show that time-delayed latent causal influences are reliably identified and that our approach considerably outperforms existing baselines that do not correctly exploit this modular representation of changes.

This article addresses the trajectory consensus problem of coordinating the trajectories of vehicles at multiple future time points. The objective is the consensus of the geometry of the vehicles' planned trajectories. The geometric feature of trajectories is parameterized by a set of trajectory states defined as required lengths along the trajectory to reduce the distance to its ending point to specific values. To solve this special consensus problem involving coupled state variables, the conventional consensus model is extended by attaching it to a mapping from the state variables to the trajectory's geometry. This mapping is established using a homotopic structure that creates a compact and efficient form for the mapping. The geometry of the homotopic structure is based on the shapes of its envelopes, and the elements in the structure are derived from their deformation. Through a homotopic search in the structure, an asymptotic consensus of trajectory states is achieved. Simulation results show the proposed coupled state consensus method can achieve better performance on the consensus of multiple vehicles than the conventional isolated state consensus method.

This paper examines a trajectory homotopy optimization framework for multiple unmanned aerial vehicles (multi-UAV) to solve the problem of dynamic penetration mission planning (PMP) with hostile obstacles and perception constraints. Constrained problems are usually more challenging and difficult to solve with some practical constraints and requirements. To improve the efficiency of the solution for the penetration path, a novel variable-time mechanism has been constructed to adapt to the updated delay time of unknown target search (UTS) and dynamic trajectory planning (DTP) two stages. The occupancy grid maps are established by a Gaussian probability field (GPF) for predicting the positions of enemy UAVs. To fully consider the hostile obstacle constraint, a hybrid adaptive obstacle avoidance approach dynamic window PRM (DW-PRM) is designed to shorten the planned path. The penetration strategy algorithm (SG) is developed based on the proposed strategy set and decision tree. To improve the ability of dynamic obstacle avoidance, the multiple coupled penetration homotopy trajectory is addressed with a turning radius constraint. The simulation results indicated that the penetration homotopy framework for multi-constraints can solve the multi-UAV PMP problem.

In the above article[1],Table I,III, andIVshould show “N/m” instead of “kN/m” and they should also show “Ns/m” instead of “kNs/m.” The revised tables are shown below.

In this paper, the traditional shortest path planning problem for vehicle is advanced to length-targeted path planning problem, i.e., to plan path with its length being as close to a specified value as possible. Lengthening and shortening of given initial paths are used to solve this problem. Based on an operation set consisting of three basic path homotopies, we build a comprehensive and systematic framework to plan paths with target length, which is a generalization of the existing related studies. Thereby, the expected paths can be independently searched through such deformation processes within topological classes. The proposed framework can produce the largest length coverage in different scenarios and under different conditions, and it can also generate expected paths with arbitrary topological classification in terms of the curvature constraint and the obstacle constraint. Examples show that our lengthening and shortening method can effectively solve the length-targeted path planning problem in environment without or with obstacles.

A novel data-driven framework to monitor the health status of high-speed rail suspension system by measuring train vibrations is proposed herein. Unlike existing methods, this framework does not rely on sophisticated dynamic models or high-fidelity simulations; it combines the power of data and domain knowledge to generate a model that can be trained quickly and adapted easily to different rail systems. In addition, the framework includes a module to generate a training dataset, tackling a typical challenge in real-world system monitoring, namely, the lack of labeled data due to practical limits. Based on the multi-output support vector regression (MSVR), the proposed framework can monitor the stiffness and damping coefficients of the suspension system using vibration signals measured on trains in real time. The framework comprises three modules. First, a simple suspension system dynamics model is built to generate a training dataset. Furthermore, key features are extracted from frequency response curves to reflect the impact of spring and damper degradation. Subsequently, a supervised learning model based on the MSVR is built to predict the stiffness and damping coefficients of suspension systems from features extracted in the second module. Once the model is built, real-time monitoring can be achieved by feeding the vibration signals as they are collected during operations. The proposed framework was evaluated on simulation data for its accuracy and tested on real-world operational data for its practicability.