Robotics and Autonomous Systems最新文献_第9页

SR-SLAM: Scene reliability-based RGB-D SLAM in diverse environments SR-SLAM：不同环境下基于场景可靠性的RGB-D SLAM

IF 5.2 2区计算机科学 Q1 AUTOMATION & CONTROL SYSTEMS

Robotics and Autonomous Systems

Pub Date : 2026-03-01 Epub Date: 2025-12-17 DOI: 10.1016/j.robot.2025.105306

Haolan Zhang , Chenghao Li , Thanh Nguyen Canh , Lijun Wang , Nak Young Chong

Visual simultaneous localization and mapping (SLAM) plays a critical role in autonomous robotic systems, especially where accurate and reliable measurements are essential for navigation and sensing. In feature-based SLAM, the quantity and quality of extracted features significantly influence system performance. Due to the variations in feature quantity and quality across diverse environments, current approaches face two major challenges: (1) limited adaptability in dynamic feature culling and pose estimation, and (2) insufficient environmental awareness in assessment and optimization strategies. To address these issues, we propose SR-SLAM, a scene reliability-based framework that enhances feature-based SLAM through environment-aware processing. Our method introduces a unified scene reliability assessment mechanism that incorporates multiple metrics and historical observations to guide system behavior. Based on this assessment, we develop: (i) adaptive dynamic region selection with flexible geometric constraints, (ii) depth-assisted self-adjusting clustering for efficient dynamic feature removal in high-dimensional settings, and (iii) reliability-aware pose refinement that dynamically integrates direct methods when features are insufficient. Furthermore, we propose (iv) reliability-based keyframe selection and a weighted optimization scheme to reduce computational overhead while improving estimation accuracy. Extensive experiments on public datasets and real-world scenarios show that SR-SLAM outperforms state-of-the-art dynamic SLAM methods, achieving up to 90% improvement in accuracy and robustness across diverse environments. These improvements directly contribute to enhanced measurement precision and reliability in autonomous robotic sensing systems.

视觉同步定位和测绘（SLAM）在自主机器人系统中起着至关重要的作用，特别是在精确可靠的测量对导航和传感至关重要的情况下。在基于特征的SLAM中，提取特征的数量和质量显著影响系统性能。由于不同环境下特征数量和质量的差异，现有方法面临两个主要挑战：(1)动态特征剔除和姿态估计的适应性有限；(2)评估和优化策略的环境意识不足。为了解决这些问题，我们提出了基于场景可靠性的SR-SLAM框架，该框架通过环境感知处理增强了基于特征的SLAM。我们的方法引入了一个统一的场景可靠性评估机制，该机制结合了多个指标和历史观察来指导系统行为。基于这一评估，我们开发了：(i)具有灵活几何约束的自适应动态区域选择，（ii）深度辅助自调整聚类，在高维环境中有效地去除动态特征，以及（iii）在特征不足时动态集成直接方法的可靠性感知姿态优化。此外，我们提出（iv）基于可靠性的关键帧选择和加权优化方案，以减少计算开销，同时提高估计精度。在公共数据集和真实场景上的大量实验表明，SR-SLAM优于最先进的动态SLAM方法，在不同环境下的准确性和鲁棒性提高了90%。这些改进直接有助于提高自主机器人传感系统的测量精度和可靠性。

{"title":"SR-SLAM: Scene reliability-based RGB-D SLAM in diverse environments","authors":"Haolan Zhang , Chenghao Li , Thanh Nguyen Canh , Lijun Wang , Nak Young Chong","doi":"10.1016/j.robot.2025.105306","DOIUrl":"10.1016/j.robot.2025.105306","url":null,"abstract":"<div><div>Visual simultaneous localization and mapping (SLAM) plays a critical role in autonomous robotic systems, especially where accurate and reliable measurements are essential for navigation and sensing. In feature-based SLAM, the quantity and quality of extracted features significantly influence system performance. Due to the variations in feature quantity and quality across diverse environments, current approaches face two major challenges: (1) limited adaptability in dynamic feature culling and pose estimation, and (2) insufficient environmental awareness in assessment and optimization strategies. To address these issues, we propose SR-SLAM, a scene reliability-based framework that enhances feature-based SLAM through environment-aware processing. Our method introduces a unified scene reliability assessment mechanism that incorporates multiple metrics and historical observations to guide system behavior. Based on this assessment, we develop: (i) adaptive dynamic region selection with flexible geometric constraints, (ii) depth-assisted self-adjusting clustering for efficient dynamic feature removal in high-dimensional settings, and (iii) reliability-aware pose refinement that dynamically integrates direct methods when features are insufficient. Furthermore, we propose (iv) reliability-based keyframe selection and a weighted optimization scheme to reduce computational overhead while improving estimation accuracy. Extensive experiments on public datasets and real-world scenarios show that SR-SLAM outperforms state-of-the-art dynamic SLAM methods, achieving up to 90% improvement in accuracy and robustness across diverse environments. These improvements directly contribute to enhanced measurement precision and reliability in autonomous robotic sensing systems.</div></div>","PeriodicalId":49592,"journal":{"name":"Robotics and Autonomous Systems","volume":"197 ","pages":"Article 105306"},"PeriodicalIF":5.2,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145840373","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

CRESTA: A Cognitivist Robot Execution framework for Semantic-driven Task Awareness 一个语义驱动任务感知的认知主义机器人执行框架

IF 5.2 2区计算机科学 Q1 AUTOMATION & CONTROL SYSTEMS

Robotics and Autonomous Systems

Pub Date : 2026-03-01 Epub Date: 2025-12-15 DOI: 10.1016/j.robot.2025.105303

Damiano Gasperini , Luca Muratore , Nikos Tsagarakis

The advancement of autonomous robots is still in need of a comprehensive framework for task execution awareness, enabling the generation of autonomous behaviors and responses in unknown dynamic environments. Situation awareness aims to enhance robot autonomy and adaptation during task execution by effectively combining key capabilities, such as reasoning, planning, projection of actions’ effects into future states, as well as perception and comprehension of the surroundings. In this work, we propose CRESTA, a novel cognitivist framework for semantic-driven task awareness that addresses the intricate challenges of perceiving, navigating, and manipulating dynamic environments. CRESTA’s objective of achieving effective robot awareness relies on the perceived environment semantics and on the combined use of online planning, reasoning, and monitoring, while also enabling recovery from task-level failures. It is designed as a set of online modules for (a) collecting and analyzing multi-sensor data as well as updating the world model description, (b) real-time decision-making and task states monitoring, and (c) execution of each action. Being highly modular and configurable to assorted robotic systems, the proposed framework aims for adaptability across diverse robotic platforms and tasks. In this work, a detailed description of CRESTA’s framework comes along with demonstrative tasks to showcase its capabilities on both the CENTAURO robot and on a custom 6 DoF manipulator. In the discussed experimental results, CRESTA leads the robot to open a door or to navigate and manipulate a lever, while recovering from failures by adapting the parameters of its actions.

自主机器人的发展仍然需要一个全面的任务执行感知框架，使其能够在未知的动态环境中产生自主行为和响应。情境感知旨在通过有效地结合关键能力，如推理、规划、将行动影响投射到未来状态，以及对周围环境的感知和理解，增强机器人在任务执行过程中的自主性和适应性。在这项工作中，我们提出了CRESTA，一个新的认知主义框架，用于语义驱动的任务意识，解决感知，导航和操纵动态环境的复杂挑战。CRESTA的目标是实现有效的机器人感知，这依赖于感知到的环境语义，以及在线规划、推理和监控的组合使用，同时还能够从任务级故障中恢复。它被设计为一组在线模块，用于(a)收集和分析多传感器数据以及更新世界模型描述，(b)实时决策和任务状态监测，以及(c)执行每个动作。该框架具有高度模块化和可配置性，可用于各种机器人系统，旨在适应各种机器人平台和任务。在这项工作中，CRESTA框架的详细描述伴随着演示任务来展示其在CENTAURO机器人和定制6自由度机械手上的功能。在讨论的实验结果中，CRESTA引导机器人打开门或导航和操纵杠杆，同时通过调整其动作参数从失败中恢复过来。

{"title":"CRESTA: A Cognitivist Robot Execution framework for Semantic-driven Task Awareness","authors":"Damiano Gasperini , Luca Muratore , Nikos Tsagarakis","doi":"10.1016/j.robot.2025.105303","DOIUrl":"10.1016/j.robot.2025.105303","url":null,"abstract":"<div><div>The advancement of autonomous robots is still in need of a comprehensive framework for task execution awareness, enabling the generation of autonomous behaviors and responses in unknown dynamic environments. Situation awareness aims to enhance robot autonomy and adaptation during task execution by effectively combining key capabilities, such as reasoning, planning, projection of actions’ effects into future states, as well as perception and comprehension of the surroundings. In this work, we propose CRESTA, a novel cognitivist framework for semantic-driven task awareness that addresses the intricate challenges of perceiving, navigating, and manipulating dynamic environments. CRESTA’s objective of achieving effective robot awareness relies on the perceived environment semantics and on the combined use of online planning, reasoning, and monitoring, while also enabling recovery from task-level failures. It is designed as a set of online modules for (a) collecting and analyzing multi-sensor data as well as updating the world model description, (b) real-time decision-making and task states monitoring, and (c) execution of each action. Being highly modular and configurable to assorted robotic systems, the proposed framework aims for adaptability across diverse robotic platforms and tasks. In this work, a detailed description of CRESTA’s framework comes along with demonstrative tasks to showcase its capabilities on both the CENTAURO robot and on a custom 6 DoF manipulator. In the discussed experimental results, CRESTA leads the robot to open a door or to navigate and manipulate a lever, while recovering from failures by adapting the parameters of its actions.</div></div>","PeriodicalId":49592,"journal":{"name":"Robotics and Autonomous Systems","volume":"197 ","pages":"Article 105303"},"PeriodicalIF":5.2,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145790308","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Passive Reinforcement Learning with Optimal Control for Safe Convergence in Cyber–physical Systems 网络物理系统安全收敛的最优控制被动强化学习

IF 5.2 2区计算机科学 Q1 AUTOMATION & CONTROL SYSTEMS

Robotics and Autonomous Systems

Pub Date : 2026-03-01 Epub Date: 2025-12-19 DOI: 10.1016/j.robot.2025.105293

Nicola Piccinelli , Daniele Meli , Enrico Bonoldi , Riccardo Muradore

Reinforcement Learning (RL) can compute optimal strategies for accomplishing difficult tasks in complex scenarios. However, most RL algorithms do not provide safety and performance guarantees during the deployment phase. This is a critical drawback when RL is applied to cyber–physical systems such as robotic manipulators, where the goal is to always and safely converge to a desired goal or equilibrium state. Specifically, one fundamental safety requirement for robotic systems is closed-loop

L_{2}

-stability, which has passivity as a sufficient condition. This paper proposes a novel switched RL control scheme for robotic systems, with passivity and asymptotic stability guarantees. This combines RL over constrained Markov decision processes, for passive training and inference, with Linear Quadratic Regulation (LQR) for asymptotic convergence to the desired equilibrium point. During RL training, the energy stored in the system is monitored via the virtual energy tank approach to train a cost critic function. During inference, the virtual energy tank modulates the command input to guarantee passivity. Finally, the reward design of the RL agent is based on the Lyapunov function associated with LQR control, in order to steer the system state towards the LQR basin of attraction, where a switching mechanism is triggered to guarantee asymptotic convergence. We compare our methodology with a model-based controller and other RL and model-based architectures applied to a paradigmatic under-actuated cart–pole system, an instance of a 2-DOF robotic manipulator, both in simulation and on a real setup. We also test the generality of our approach, with an experiment on a 6-DOF manipulator in simulation. The experimental validation shows that our methodology performs better in training and inference, even in the presence of plant modelling errors, while guaranteeing passivity and safety in the presence of large disruptive disturbances.

强化学习（RL）可以计算出在复杂场景中完成困难任务的最佳策略。然而，大多数强化学习算法在部署阶段不提供安全和性能保证。当RL应用于网络物理系统（如机器人操纵器）时，这是一个关键的缺点，其目标是始终安全地收敛到期望的目标或平衡状态。其中，以无源性为充分条件的闭环l2稳定性是机器人系统安全的一个基本要求。提出了一种具有无源性和渐近稳定性保证的机器人系统切换RL控制方案。这结合了RL的约束马尔可夫决策过程，被动训练和推理，线性二次调节（LQR）渐近收敛到期望的平衡点。在RL训练过程中，通过虚拟能量罐方法监测系统中存储的能量，以训练成本批判函数。在推理过程中，虚拟能量罐对命令输入进行调制，保证无源性。最后，RL智能体的奖励设计基于与LQR控制相关联的Lyapunov函数，目的是将系统状态引导到LQR吸引池，在LQR吸引池中触发切换机制以保证渐近收敛。我们将我们的方法与基于模型的控制器和其他基于RL和基于模型的体系结构进行了比较，这些体系结构应用于典型的欠驱动推车杆系统，这是一个2自由度机器人操纵器的实例，无论是在模拟中还是在实际设置中。我们还通过一个六自由度机械臂的仿真实验验证了我们方法的通用性。实验验证表明，即使在存在植物建模错误的情况下，我们的方法在训练和推理方面表现更好，同时保证了在存在大型破坏性干扰的情况下的被动性和安全性。

{"title":"Passive Reinforcement Learning with Optimal Control for Safe Convergence in Cyber–physical Systems","authors":"Nicola Piccinelli , Daniele Meli , Enrico Bonoldi , Riccardo Muradore","doi":"10.1016/j.robot.2025.105293","DOIUrl":"10.1016/j.robot.2025.105293","url":null,"abstract":"<div><div>Reinforcement Learning (RL) can compute optimal strategies for accomplishing difficult tasks in complex scenarios. However, most RL algorithms do not provide safety and performance guarantees during the deployment phase. This is a critical drawback when RL is applied to cyber–physical systems such as robotic manipulators, where the goal is to always and safely converge to a desired goal or equilibrium state. Specifically, one fundamental safety requirement for robotic systems is closed-loop <span><math><msub><mrow><mi>L</mi></mrow><mrow><mn>2</mn></mrow></msub></math></span>-stability, which has <em>passivity</em> as a sufficient condition. This paper proposes a novel switched RL control scheme for robotic systems, with <em>passivity and asymptotic stability guarantees</em>. This combines RL over constrained Markov decision processes, for <em>passive training and inference</em>, with Linear Quadratic Regulation (LQR) for asymptotic convergence to the desired equilibrium point. During RL training, the energy stored in the system is monitored via the virtual energy tank approach to train a cost critic function. During inference, the virtual energy tank modulates the command input to guarantee passivity. Finally, the reward design of the RL agent is based on the Lyapunov function associated with LQR control, in order to steer the system state towards the LQR basin of attraction, where a switching mechanism is triggered to guarantee asymptotic convergence. We compare our methodology with a model-based controller and other RL and model-based architectures applied to a paradigmatic under-actuated cart–pole system, an instance of a 2-DOF robotic manipulator, both in simulation and on a real setup. We also test the generality of our approach, with an experiment on a 6-DOF manipulator in simulation. The experimental validation shows that our methodology performs better in training and inference, even in the presence of plant modelling errors, while guaranteeing passivity and safety in the presence of large disruptive disturbances.</div></div>","PeriodicalId":49592,"journal":{"name":"Robotics and Autonomous Systems","volume":"197 ","pages":"Article 105293"},"PeriodicalIF":5.2,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145790831","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Beyond coverage path planning: Can UAV swarms perfect scattered regions inspections? 超越覆盖路径规划：无人机群能否完善分散区域巡检？

IF 5.2 2区计算机科学 Q1 AUTOMATION & CONTROL SYSTEMS

Robotics and Autonomous Systems

Pub Date : 2026-03-01 Epub Date: 2025-12-10 DOI: 10.1016/j.robot.2025.105297

Socratis Gkelios , Savvas D. Apostolidis , Pavlos Ch. Kapoutsis , Elias B. Kosmatopoulos , Athanasios Ch. Kapoutsis

Unmanned Aerial Vehicles (UAVs) have revolutionized inspection tasks by offering a safer, more efficient, and flexible alternative to traditional methods. However, battery limitations often constrain their effectiveness, necessitating the development of optimized flight paths and data collection techniques. While existing approaches like coverage path planning (CPP) ensure comprehensive data collection, they can be inefficient, especially when inspecting multiple non-connected Regions of Interest (ROIs). This paper introduces the Fast Inspection of Scattered Regions (FISR) problem and proposes a novel solution, the multi-UAV Disjoint Areas Inspection (mUDAI) method. The introduced approach implements a two-fold optimization procedure, for calculating the best image capturing positions and the most efficient UAV trajectories, balancing data resolution and operational time, minimizing redundant data collection and resource consumption. The mUDAI method is designed to enable rapid, efficient inspections of scattered ROIs, making it ideal for applications such as security infrastructure assessments, agricultural inspections, and emergency site evaluations. A combination of simulated evaluations and real-world deployments is used to validate and quantify the method’s ability to improve operational efficiency while preserving high-quality data capture, demonstrating its effectiveness in real-world operations. An open-source Python implementation of the mUDAI method can be found on GitHub¹ and the collected and processed data from the real-world experiments are all hosted on Zenodo². Finally, this on-line platform³ allows the interested readers to interact with the mUDAI method and generate their own multi-UAV FISR missions.

无人驾驶飞行器（uav）通过提供一种更安全、更高效、更灵活的替代传统方法，彻底改变了检查任务。然而，电池的限制往往限制了它们的有效性，这就需要开发优化的飞行路径和数据收集技术。虽然覆盖路径规划（CPP）等现有方法可以确保全面的数据收集，但它们可能效率低下，特别是在检查多个未连接的感兴趣区域（roi）时。介绍了离散区域快速检测（FISR）问题，提出了一种新的解决方案——多无人机离散区域检测（mUDAI）方法。引入的方法实现了双重优化程序，用于计算最佳图像捕获位置和最有效的无人机轨迹，平衡数据分辨率和操作时间，最大限度地减少冗余数据收集和资源消耗。mUDAI方法旨在对分散的roi进行快速、有效的检查，使其成为安全基础设施评估、农业检查和紧急现场评估等应用的理想选择。模拟评估和实际部署相结合，用于验证和量化该方法提高操作效率的能力，同时保持高质量的数据捕获，证明其在实际操作中的有效性。mUDAI方法的开源Python实现可以在GitHub1上找到，从真实世界的实验中收集和处理的数据都托管在Zenodo2上。最后，这个在线平台3允许感兴趣的读者与mUDAI方法进行交互，并生成自己的多无人机FISR任务。

{"title":"Beyond coverage path planning: Can UAV swarms perfect scattered regions inspections?","authors":"Socratis Gkelios , Savvas D. Apostolidis , Pavlos Ch. Kapoutsis , Elias B. Kosmatopoulos , Athanasios Ch. Kapoutsis","doi":"10.1016/j.robot.2025.105297","DOIUrl":"10.1016/j.robot.2025.105297","url":null,"abstract":"<div><div>Unmanned Aerial Vehicles (UAVs) have revolutionized inspection tasks by offering a safer, more efficient, and flexible alternative to traditional methods. However, battery limitations often constrain their effectiveness, necessitating the development of optimized flight paths and data collection techniques. While existing approaches like coverage path planning (CPP) ensure comprehensive data collection, they can be inefficient, especially when inspecting multiple non-connected Regions of Interest (ROIs). This paper introduces the Fast Inspection of Scattered Regions (FISR) problem and proposes a novel solution, the multi-UAV Disjoint Areas Inspection (mUDAI) method. The introduced approach implements a two-fold optimization procedure, for calculating the best image capturing positions and the most efficient UAV trajectories, balancing data resolution and operational time, minimizing redundant data collection and resource consumption. The mUDAI method is designed to enable rapid, efficient inspections of scattered ROIs, making it ideal for applications such as security infrastructure assessments, agricultural inspections, and emergency site evaluations. A combination of simulated evaluations and real-world deployments is used to validate and quantify the method’s ability to improve operational efficiency while preserving high-quality data capture, demonstrating its effectiveness in real-world operations. An open-source Python implementation of the mUDAI method can be found on GitHub<span><span><sup>1</sup></span></span> and the collected and processed data from the real-world experiments are all hosted on Zenodo<span><span><sup>2</sup></span></span>. Finally, this on-line platform<span><span><sup>3</sup></span></span> allows the interested readers to interact with the mUDAI method and generate their own multi-UAV FISR missions.</div></div>","PeriodicalId":49592,"journal":{"name":"Robotics and Autonomous Systems","volume":"197 ","pages":"Article 105297"},"PeriodicalIF":5.2,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145790832","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Multi-goal path planning for robot-aided transcranial magnetic stimulation 机器人辅助经颅磁刺激的多目标路径规划

IF 5.2 2区计算机科学 Q1 AUTOMATION & CONTROL SYSTEMS

Robotics and Autonomous Systems

Pub Date : 2026-03-01 Epub Date: 2025-12-11 DOI: 10.1016/j.robot.2025.105294

Lixia Wang , Qing Tang , Haoyang Xing , Jiarui Dong

Since the proposal of transcranial magnetic stimulation (TMS), it has found extensive applications in the treatment of brain disorders and other fields. Robot-assisted TMS therapy ensures the accuracy and durability of treatment positioning. However, the use of robot-assisted therapy has also raised potential collisions and path-planning challenges. Due to the redundant degree of freedom (DOF), the robot has infinite configurations while performing each given stimulation. The problem that involves visiting multiple target points and returning to the initial position in TMS therapy is a combination of a generalized traveling salesman problem with neighborhoods (GTSPN) and a collision-free path planning problem. A global method based on the probabilistic roadmap (PRM) and the A* algorithm is proposed. To safely visit target points, the proposed method samples uniformly on a sphere. A special nearest neighbor definition and delayed collision queries for edges are introduced to accelerate the roadmap construction. Then, the proposed method utilizes a modified A* algorithm with a multi-goal heuristic to rapidly search for a global path through all target points. Finally, a strategy combining local re-search and global re-search is proposed to get the final collision-free path. Experiments are conducted on both simulation and physical platforms using a typical model of the TMS therapy system. The results indicate that our proposed algorithm can effectively avoid collisions and produce optimal planning results for TMS therapy in a short time.

经颅磁刺激（transcranial magnetic stimulation， TMS）自提出以来，在脑部疾病的治疗等领域得到了广泛的应用。机器人辅助TMS治疗确保了治疗定位的准确性和耐久性。然而，机器人辅助疗法的使用也带来了潜在的碰撞和路径规划方面的挑战。由于冗余自由度（DOF）的存在，机器人在执行每次给定的激励时具有无限种构型。在TMS治疗中，涉及到访问多个目标点并返回初始位置的问题是一个带邻域的广义旅行商问题（GTSPN）和无碰撞路径规划问题的结合。提出了一种基于概率路线图（PRM）和A*算法的全局方法。为了安全访问目标点，该方法在球体上均匀采样。引入了一种特殊的最近邻定义和边的延迟碰撞查询来加速路线图的构建。然后，利用改进的a *算法和多目标启发式算法，通过所有目标点快速搜索全局路径。最后，提出局部研究与全局研究相结合的策略，得到最终的无碰撞路径。利用典型的经颅磁刺激治疗系统模型，在仿真和物理平台上进行了实验。结果表明，本文提出的算法可以有效避免碰撞，并在短时间内为TMS治疗提供最优规划结果。

{"title":"Multi-goal path planning for robot-aided transcranial magnetic stimulation","authors":"Lixia Wang , Qing Tang , Haoyang Xing , Jiarui Dong","doi":"10.1016/j.robot.2025.105294","DOIUrl":"10.1016/j.robot.2025.105294","url":null,"abstract":"<div><div>Since the proposal of transcranial magnetic stimulation (TMS), it has found extensive applications in the treatment of brain disorders and other fields. Robot-assisted TMS therapy ensures the accuracy and durability of treatment positioning. However, the use of robot-assisted therapy has also raised potential collisions and path-planning challenges. Due to the redundant degree of freedom (DOF), the robot has infinite configurations while performing each given stimulation. The problem that involves visiting multiple target points and returning to the initial position in TMS therapy is a combination of a generalized traveling salesman problem with neighborhoods (GTSPN) and a collision-free path planning problem. A global method based on the probabilistic roadmap (PRM) and the A* algorithm is proposed. To safely visit target points, the proposed method samples uniformly on a sphere. A special nearest neighbor definition and delayed collision queries for edges are introduced to accelerate the roadmap construction. Then, the proposed method utilizes a modified A* algorithm with a multi-goal heuristic to rapidly search for a global path through all target points. Finally, a strategy combining local re-search and global re-search is proposed to get the final collision-free path. Experiments are conducted on both simulation and physical platforms using a typical model of the TMS therapy system. The results indicate that our proposed algorithm can effectively avoid collisions and produce optimal planning results for TMS therapy in a short time.</div></div>","PeriodicalId":49592,"journal":{"name":"Robotics and Autonomous Systems","volume":"197 ","pages":"Article 105294"},"PeriodicalIF":5.2,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145790834","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Decentralized formation control, collision avoidance and global connectivity maintenance using non-smooth barrier functions 分散式编队控制、避碰和使用非平滑屏障函数的全局连通性维护

IF 5.2 2区计算机科学 Q1 AUTOMATION & CONTROL SYSTEMS

Robotics and Autonomous Systems

Pub Date : 2026-03-01 Epub Date: 2025-12-01 DOI: 10.1016/j.robot.2025.105272

Pranjal Bhatia , Sayan Basu Roy , P.B. Sujit , Luis Mejias Alvarez , Aaron McFadyen

This paper presents a decentralized control framework for multi-agent systems that simultaneously addresses formation control, global connectivity maintenance, and collision avoidance. To achieve these multiple objectives, we employ control barrier functions (CBFs), with each function tailored to a specific safety constraint. However, the combination of multiple CBFs inherently introduces non-smoothness in the system, posing challenges for real-time control. To address this, we adopt non-smooth barrier functions (NBFs), which enable the integration of multiple safety constraints in a decentralized manner while systematically avoiding critical non-smooth regions. The proposed method allows agents to maintain a desired formation, avoid inter-agent and obstacle collisions, and preserve global connectivity without requiring centralized coordination. The results are validated using mathematical analysis and simulations.

本文提出了一个多智能体系统的分散控制框架，该框架同时解决了编队控制、全局连接维护和避免碰撞的问题。为了实现这些多重目标，我们采用了控制屏障函数（cbf），每个函数都针对特定的安全约束进行了定制。然而，多个cbf的组合固有地引入了系统的非平滑性，给实时控制带来了挑战。为了解决这个问题，我们采用了非光滑屏障函数（nbf），它可以以分散的方式集成多个安全约束，同时系统地避免关键的非光滑区域。该方法允许智能体在不需要集中协调的情况下保持理想的队形，避免智能体间和障碍物碰撞，并保持全局连通性。通过数学分析和仿真验证了结果。

引用次数: 0

AUV decision-making in bistatic sonar target tracking via deep reinforcement learning 基于深度强化学习的双基地声纳目标跟踪中的AUV决策

IF 5.2 2区计算机科学 Q1 AUTOMATION & CONTROL SYSTEMS

Robotics and Autonomous Systems

Pub Date : 2026-03-01 Epub Date: 2025-11-26 DOI: 10.1016/j.robot.2025.105262

Weicong Zhan , Yu Tian , Feng Zheng , Qiming Sang , Jiancheng Yu

The maneuvering decisions of an Autonomous Underwater Vehicle (AUV) equipped with a sonar receiver are critical for acquiring informative measurements, thereby enhancing target tracking accuracy and maintaining persistent track continuity. This study investigates a bistatic sonar configuration comprising a stationary acoustic source and a mobile AUV carrying a linear sonar array, with a focus on real-time generation of optimized heading commands for the AUV. To address the practical limitations of conventional model-based strategies under stringent decision-time constraints, a deep reinforcement learning (DRL) framework is employed to derive an efficient decision policy. Implemented using the Proximal Policy Optimization algorithm, the DRL policy integrates with a target tracker by utilizing estimated target states and associated uncertainties as inputs. A reward function based on the Optimal Sub-Pattern Assignment metric is designed to guide policy learning. To train the policy, a comprehensive simulation environment is developed, incorporating target properties, acoustic propagation characteristics, and bistatic measurement models. Once trained, the DRL-based policy effectively guides the AUV in real-time to establish favorable bistatic configurations and enhance tracking performance. Comparative Monte Carlo simulations against a state-of-the-art model-based receding horizon control method demonstrate that the DRL-based approach yields a 14.5 % average improvement in tracking accuracy. Detailed analyses of AUV behaviors under the learned policy, along with robustness and generalization evaluations, underscore the practical potential of the proposed approach for real-world deployment in bistatic sonar target tracking scenarios.

装备声呐接收器的自主水下航行器（AUV）的机动决策对于获取信息测量，从而提高目标跟踪精度和保持持久的跟踪连续性至关重要。本研究研究了一种双基地声纳配置，包括一个固定声源和一个携带线性声纳阵列的移动AUV，重点是为AUV实时生成优化的航向命令。为了解决传统的基于模型的策略在严格的决策时间约束下的实际局限性，采用深度强化学习（DRL）框架来推导有效的决策策略。DRL策略使用近端策略优化算法实现，通过利用估计的目标状态和相关的不确定性作为输入，与目标跟踪器集成。设计了一个基于最优子模式分配度量的奖励函数来指导策略学习。为了训练该策略，开发了一个综合仿真环境，包括目标特性、声传播特性和双基地测量模型。经过训练后，基于drl的策略可以有效地实时引导AUV建立有利的双基地配置，提高跟踪性能。蒙特卡罗仿真与最先进的基于模型的后退地平线控制方法的对比表明，基于drl的方法在跟踪精度方面平均提高了14.5%。对学习策略下AUV行为的详细分析，以及鲁棒性和泛化评估，强调了该方法在双基地声纳目标跟踪场景中实际部署的实际潜力。

{"title":"AUV decision-making in bistatic sonar target tracking via deep reinforcement learning","authors":"Weicong Zhan , Yu Tian , Feng Zheng , Qiming Sang , Jiancheng Yu","doi":"10.1016/j.robot.2025.105262","DOIUrl":"10.1016/j.robot.2025.105262","url":null,"abstract":"<div><div>The maneuvering decisions of an Autonomous Underwater Vehicle (AUV) equipped with a sonar receiver are critical for acquiring informative measurements, thereby enhancing target tracking accuracy and maintaining persistent track continuity. This study investigates a bistatic sonar configuration comprising a stationary acoustic source and a mobile AUV carrying a linear sonar array, with a focus on real-time generation of optimized heading commands for the AUV. To address the practical limitations of conventional model-based strategies under stringent decision-time constraints, a deep reinforcement learning (DRL) framework is employed to derive an efficient decision policy. Implemented using the Proximal Policy Optimization algorithm, the DRL policy integrates with a target tracker by utilizing estimated target states and associated uncertainties as inputs. A reward function based on the Optimal Sub-Pattern Assignment metric is designed to guide policy learning. To train the policy, a comprehensive simulation environment is developed, incorporating target properties, acoustic propagation characteristics, and bistatic measurement models. Once trained, the DRL-based policy effectively guides the AUV in real-time to establish favorable bistatic configurations and enhance tracking performance. Comparative Monte Carlo simulations against a state-of-the-art model-based receding horizon control method demonstrate that the DRL-based approach yields a 14.5 % average improvement in tracking accuracy. Detailed analyses of AUV behaviors under the learned policy, along with robustness and generalization evaluations, underscore the practical potential of the proposed approach for real-world deployment in bistatic sonar target tracking scenarios.</div></div>","PeriodicalId":49592,"journal":{"name":"Robotics and Autonomous Systems","volume":"197 ","pages":"Article 105262"},"PeriodicalIF":5.2,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145646035","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

The FriûlBot dataset: Experimental validation of an autonomous ground robot for vineyard 3D mapping fri<s:1> lbot数据集：葡萄园3D测绘自主地面机器人的实验验证

IF 5.2 2区计算机科学 Q1 AUTOMATION & CONTROL SYSTEMS

Robotics and Autonomous Systems

Pub Date : 2026-03-01 Epub Date: 2025-12-05 DOI: 10.1016/j.robot.2025.105283

Diego Tiozzo Fasiolo, Lorenzo Scalera, Eleonora Maset, Alessandro Gasparetto

In this paper, we present the FriûlBot open-source dataset, collected by an autonomous mobile robot for vineyard 3D mapping and monitoring. The dataset is designed to provide a reference testbed for the development and benchmarking of localization, mapping, and multi-sensor data fusion algorithms. It includes detailed information on the robot status, point clouds of the vineyard, and multispectral images, offering valuable resources for future research on autonomous robotic systems in agriculture. The mobile robot employed for the dataset acquisition is capable of autonomously navigating, reaching GNSS way points, and building a map of the canopy integrating geometric and multispectral data. The navigation and mapping approach proposed in this work is based on a SLAM algorithm that integrates multiple odometry sources, and is designed for agricultural environments with plants arranged in rows, e.g., vineyards and orchards. The performance of the proposed mobile robot and navigation approach are tested during an extensive experimental campaign in a vineyard of University of Udine (Italy). During the tests, the robot successfully navigates along several vineyard rows, building point clouds of the environment that are merged with data regarding multiple vegetation indexes. The experimental results confirm the reliability of the autonomous mobile robot and the potential of the proposed dataset to foster further advances in robotics for vineyard 3D mapping.

在本文中，我们展示了fri lbot开源数据集，该数据集由自主移动机器人收集，用于葡萄园3D测绘和监测。该数据集旨在为定位、制图和多传感器数据融合算法的开发和基准测试提供参考测试平台。它包括机器人状态的详细信息，葡萄园的点云和多光谱图像，为未来农业自主机器人系统的研究提供了宝贵的资源。用于数据采集的移动机器人能够自主导航，到达GNSS路径点，并结合几何和多光谱数据构建冠层地图。本文提出的导航和制图方法基于SLAM算法，该算法集成了多个里程计源，适用于植物成行排列的农业环境，例如葡萄园和果园。所提出的移动机器人和导航方法的性能在乌迪内大学（意大利）葡萄园的广泛实验活动中进行了测试。在测试过程中，机器人成功地沿着几行葡萄园进行导航，建立了与多个植被指数数据合并的环境点云。实验结果证实了自主移动机器人的可靠性，以及所提出的数据集的潜力，以促进葡萄园3D测绘机器人技术的进一步发展。

{"title":"The FriûlBot dataset: Experimental validation of an autonomous ground robot for vineyard 3D mapping","authors":"Diego Tiozzo Fasiolo, Lorenzo Scalera, Eleonora Maset, Alessandro Gasparetto","doi":"10.1016/j.robot.2025.105283","DOIUrl":"10.1016/j.robot.2025.105283","url":null,"abstract":"<div><div>In this paper, we present the FriûlBot open-source dataset, collected by an autonomous mobile robot for vineyard 3D mapping and monitoring. The dataset is designed to provide a reference testbed for the development and benchmarking of localization, mapping, and multi-sensor data fusion algorithms. It includes detailed information on the robot status, point clouds of the vineyard, and multispectral images, offering valuable resources for future research on autonomous robotic systems in agriculture. The mobile robot employed for the dataset acquisition is capable of autonomously navigating, reaching GNSS way points, and building a map of the canopy integrating geometric and multispectral data. The navigation and mapping approach proposed in this work is based on a SLAM algorithm that integrates multiple odometry sources, and is designed for agricultural environments with plants arranged in rows, e.g., vineyards and orchards. The performance of the proposed mobile robot and navigation approach are tested during an extensive experimental campaign in a vineyard of University of Udine (Italy). During the tests, the robot successfully navigates along several vineyard rows, building point clouds of the environment that are merged with data regarding multiple vegetation indexes. The experimental results confirm the reliability of the autonomous mobile robot and the potential of the proposed dataset to foster further advances in robotics for vineyard 3D mapping.</div></div>","PeriodicalId":49592,"journal":{"name":"Robotics and Autonomous Systems","volume":"197 ","pages":"Article 105283"},"PeriodicalIF":5.2,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145738115","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A survey on theories and applications for multi-robot cooperative hunting 多机器人协同狩猎理论与应用综述

IF 5.2 2区计算机科学 Q1 AUTOMATION & CONTROL SYSTEMS

Robotics and Autonomous Systems

Pub Date : 2026-03-01 Epub Date: 2025-12-11 DOI: 10.1016/j.robot.2025.105296

Jianjun Ni , Yonghao Zhao , Ziru Zhang , Chunyan Ke , Simon X. Yang

Multi-robot cooperative hunting problem has gained widespread attention due to its potential applications in fields such as surveillance, search and rescue, and military. Recently, many excellent solutions have been proposed in this field. Thus, this paper provides a comprehensive review of the latest advancements and challenges in the field of multi-robot cooperative hunting. This survey provides an overview of the key elements in this field, such as robot types, team compositions, system architectures, communication mechanisms, cooperative modes, and so on. In addition, the key issues in multi-robot cooperative hunting are analyzed, including perception and environmental understanding, task allocation, formation control, path planning, trajectory tracking, and decision-making. Then, some representative theories and methods for these issues are summarized. Finally, some potential solutions are discussed and future research directions in this field are presented.

多机器人协同狩猎问题因其在监视、搜救和军事等领域的潜在应用而受到广泛关注。近年来，在这一领域提出了许多优秀的解决方案。因此，本文对多机器人协同狩猎领域的最新进展和挑战进行了全面的综述。该调查提供了该领域关键要素的概述，例如机器人类型、团队组成、系统架构、通信机制、合作模式等。此外，分析了多机器人协同狩猎中的关键问题，包括感知与环境理解、任务分配、编队控制、路径规划、轨迹跟踪和决策。然后，对这些问题的代表性理论和方法进行了总结。最后，对可能的解决方案进行了讨论，并提出了该领域未来的研究方向。

引用次数: 0

Multi-objective optimization for dimensional synthesis of tendon placement and structural design for energy-efficient and feasible static workspace in continuum robots 面向连续体机器人节能可行静态工作空间的肌腱布置尺寸综合和结构设计多目标优化

IF 5.2 2区计算机科学 Q1 AUTOMATION & CONTROL SYSTEMS

Robotics and Autonomous Systems

Pub Date : 2026-03-01 Epub Date: 2025-12-11 DOI: 10.1016/j.robot.2025.105300

Mohammad Jabari , Carmen Visconte , Giuseppe Quaglia , Med Amine Laribi

Tendon-driven continuum robots (TDCRs) face a critical trade-off between energy efficiency and static performance for navigating constrained environments, a challenge in medical and industrial applications. This study proposes a bi-objective optimization framework to enhance tendon placement and dimensional synthesis in a two-segment TDCR, featuring seven disks and four tendons per segment. Leveraging a kineto-static model based on piecewise constant curvature (PCC) theory and a multi-objective genetic algorithm (MOGA), radial tendon distances and angular offsets have been optimized. These solutions achieve up to 30 % reduction in mechanical work and a 3–5 % workspace expansion, validated through 100 randomized tendon force samples. The results offer practical guidelines for improving TDCR performance in both minimally invasive surgery and industrial inspection.

肌腱驱动连续体机器人（tdcr）面临着能源效率和静态性能之间的关键权衡，以导航受限环境，这是医疗和工业应用中的一个挑战。本研究提出了一个双目标优化框架，以增强两节段TDCR的肌腱放置和尺寸合成，每节段有7个椎间盘和4个肌腱。利用基于分段常曲率（PCC）理论的动静态模型和多目标遗传算法（MOGA），优化了径向肌腱距离和角偏移量。这些解决方案减少了30%的机械工作量，并扩大了3 - 5%的工作空间，通过100个随机肌腱力样本进行了验证。结果为提高TDCR在微创手术和工业检查中的性能提供了实用指导。

引用次数: 0