IEEE Robotics and Automation Letters最新文献_第4页

Seeing Through Pixel Motion: Learning Obstacle Avoidance From Optical Flow With One Camera 透过像素运动：学习避障从光流与一个相机

IF 4.6 2区计算机科学

IEEE Robotics and Automation Letters Pub Date : 2025-04-15 DOI: 10.1109/LRA.2025.3560842

Yu Hu;Yuang Zhang;Yunlong Song;Yang Deng;Feng Yu;Linzuo Zhang;Weiyao Lin;Danping Zou;Wenxian Yu

{"title":"Seeing Through Pixel Motion: Learning Obstacle Avoidance From Optical Flow With One Camera","authors":"Yu Hu;Yuang Zhang;Yunlong Song;Yang Deng;Feng Yu;Linzuo Zhang;Weiyao Lin;Danping Zou;Wenxian Yu","doi":"10.1109/LRA.2025.3560842","DOIUrl":"https://doi.org/10.1109/LRA.2025.3560842","url":null,"abstract":"Optical flow captures the motion of pixels in an image sequence over time, providing information about movement, depth, and environmental structure. Flying insects utilize this information to navigate and avoid obstacles, allowing them to execute highly agile maneuvers even in complex environments. Despite its potential, autonomous flying robots have yet to fully leverage this motion information to achieve comparable levels of agility and robustness. The main challenges are two-fold: 1) extracting accurate optical flow from visual data during high-speed flight and 2) designing a robust controller that can handle noisy optical flow estimations while ensuring robust performance in complex environments. To address these challenges, we propose a novel end-to-end system for quadrotor obstacle avoidance using monocular optical flow. We develop an efficient differentiable simulator coupled with a simplified quadrotor model, allowing our policy to be trained directly through first-order gradient optimization. Additionally, we introduce a central flow attention mechanism and an action-guided active sensing strategy that enhances the policy's focus on task-relevant optical flow observations to enable more responsive decision-making during flight. Our system is validated both in simulation and the real world using an FPV racing drone. Despite being trained in a simple environment in simulation, our system demonstrates agile and robust flight in various unknown, cluttered environments in the real world at speeds of up to 6 m/s.","PeriodicalId":13241,"journal":{"name":"IEEE Robotics and Automation Letters","volume":"10 6","pages":"5871-5878"},"PeriodicalIF":4.6,"publicationDate":"2025-04-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143902625","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

Integrating Reinforcement Learning and Virtual Fixtures for Safer Automatic Robotic Surgery 集成强化学习和虚拟夹具的更安全自动机器人手术

IF 4.6 2区计算机科学

IEEE Robotics and Automation Letters Pub Date : 2025-04-15 DOI: 10.1109/LRA.2025.3559826

Ke Fan;Ziyang Chen

{"title":"Integrating Reinforcement Learning and Virtual Fixtures for Safer Automatic Robotic Surgery","authors":"Ke Fan;Ziyang Chen","doi":"10.1109/LRA.2025.3559826","DOIUrl":"https://doi.org/10.1109/LRA.2025.3559826","url":null,"abstract":"A primary concern in robotic automation is safety, especially in surgical scenarios. In this letter, we propose a virtual fixture (VF) based safe reinforcement learning framework to ensure safety constraints. The framework ensures that the agent, particularly multi-joint robotic manipulator agents, acts within the hard constraints. In the training phase, VF confines the exploration of the agent within a safe operational space. The core idea is that once the agent violates the VF, it will be pushed back to the safe region. Then, the safe action corrected by the VF is collected and forms a safe experience used for subsequent policy optimization, which we refer to as safety experience reshaping (SER). Subsequently, we design a visual module to detect safety constraints to construct the VF and transfer the trained policy to the real robot. We compare our framework to 5 state-of-the-art RL methods and a nonlearning-based method. Results show that our framework gets a lower rate of constraint violations and better performance in task success. Furthermore, in addition to the static constraint tasks, we also designed two tasks involving dynamic constraints, highlighting the superiority of our method in handling dynamic constraints. The videos of our physical experiment can be found in the following links (Lymph node removal, Human-robot collaboration 1, Human-robot collaboration 2).","PeriodicalId":13241,"journal":{"name":"IEEE Robotics and Automation Letters","volume":"10 6","pages":"5265-5272"},"PeriodicalIF":4.6,"publicationDate":"2025-04-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143848902","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

Gazing Preference Induced Controllable Milling Behavior in Swarm Robotics 注视偏好诱导的群体机器人可控铣削行为

IF 4.6 2区计算机科学

IEEE Robotics and Automation Letters Pub Date : 2025-04-15 DOI: 10.1109/LRA.2025.3560897

Yongjian Zhou;Jintao Song;Tong Liu;Xingguang Peng

{"title":"Gazing Preference Induced Controllable Milling Behavior in Swarm Robotics","authors":"Yongjian Zhou;Jintao Song;Tong Liu;Xingguang Peng","doi":"10.1109/LRA.2025.3560897","DOIUrl":"https://doi.org/10.1109/LRA.2025.3560897","url":null,"abstract":"Milling is a collective behavior that is useful in a variety of real scenarios, but how to regulate milling behavior simply and efficiently is still a challenging problem. This letter introduces a novel method for controlling milling behavior in real-world robots, where both the direction and radius of the milling pattern can be continuously adjusted by tuning a single parameter. Inspired by visual attention mechanisms, the proposed model introduces the concept of gazing preference. That is, the robot will prefer to choose a neighbor in a particular direction for interaction, which in turn creates a force that deviates from the direction of velocity and leads to the milling behavior. Additionally, a potential function ensures swarm cohesion and prevents collisions through simulated repulsive and attractive forces. Simulations and experiments involving up to 50 robots demonstrate that adjusting the gazing preference parameter enables seamless control of the rotation direction and fine-tuning of the milling pattern's angular velocity and radius. Overall, this letter provides a straightforward and effective approach for designing controlled milling behavior in swarm robotics.","PeriodicalId":13241,"journal":{"name":"IEEE Robotics and Automation Letters","volume":"10 6","pages":"5769-5776"},"PeriodicalIF":4.6,"publicationDate":"2025-04-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143900637","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

Depth Completion With Super-Resolution and Cross-Modality Optimization 具有超分辨率和跨模态优化的深度补全

IF 4.6 2区计算机科学

IEEE Robotics and Automation Letters Pub Date : 2025-04-15 DOI: 10.1109/LRA.2025.3560860

Ju Zhong;Aimin Jiang;Chang Liu;Ning Xu;Yanping Zhu

{"title":"Depth Completion With Super-Resolution and Cross-Modality Optimization","authors":"Ju Zhong;Aimin Jiang;Chang Liu;Ning Xu;Yanping Zhu","doi":"10.1109/LRA.2025.3560860","DOIUrl":"https://doi.org/10.1109/LRA.2025.3560860","url":null,"abstract":"Depth completion, the process of generating dense depth maps from sparse or incomplete data, is inherently challenging due to the variability in sensor types, environmental conditions, and the differences between data modalities. In this letter, we propose a novel framework for depth completion. First, we introduce an efficient depth estimation network capable of predicting relative depth from a single RGB image. Next, we design a depth super-resolution network that refines the predicted depth by using Fast Fourier Convolution (FFC) and Gradient-weighted Symmetric Feature Transmission (GSFT) modules. These modules upsample the depth map using high-resolution RGB guidance, effectively mitigating the cross-modality gap. Finally, a global optimization step fuses the upsampled depth with sparse ground truth to produce high-quality dense depth maps. Our unified approach enhances generalization across diverse datasets while avoiding overfitting to specific depth corruption patterns. The enhancement in depth resolution and accuracy are critical for robotic applications requiring precise spatial perception, such as localization and manipulation. Experimental results on NYU-Depth V2 and SUN RGB-D benchmarks demonstrate the superiority of our approach compared to state-of-the-art approaches.","PeriodicalId":13241,"journal":{"name":"IEEE Robotics and Automation Letters","volume":"10 6","pages":"5585-5592"},"PeriodicalIF":4.6,"publicationDate":"2025-04-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143888293","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

Motion Planning and Compensation Approaches for Autonomous Surface Manipulator Systems in Grasping Tasks on Water Surfaces 自主表面机械臂系统在水面抓取任务中的运动规划与补偿方法

IF 4.6 2区计算机科学

IEEE Robotics and Automation Letters Pub Date : 2025-04-15 DOI: 10.1109/LRA.2025.3560866

Tao Li;Yi Cai

{"title":"Motion Planning and Compensation Approaches for Autonomous Surface Manipulator Systems in Grasping Tasks on Water Surfaces","authors":"Tao Li;Yi Cai","doi":"10.1109/LRA.2025.3560866","DOIUrl":"https://doi.org/10.1109/LRA.2025.3560866","url":null,"abstract":"Autonomous surface manipulation systems (ASMSs) are novel robotics platforms composed of unmanned surface vehicles (USVs) and manipulators, and they can be used to recover floating objects on water surfaces. However, the improper positional relationship between the target object and the USV, and the dynamic coupling between the manipulator and the USV, may significantly reduce the grasping performance and stability of ASMS. To address this issue, this study proposes motion planning and compensation methods for ASMS. Firstly, an ASMS model considering the dynamic coupling between the USV and the manipulator is developed. Subsequently, a unified index for evaluating the positional relationship between the USV and target objects is introduced, and a modified visual servo control method is proposed to be used in manipulators with joint angles as motion commands. Motion compensation of the ASMS is achieved through a neural network model describing the motion coupling in ASMS. The hardware-in-the-loop simulation results illustrate that when the target object is located in regions with high values of the unified index, the grasping performance and stability of ASMS can be enhanced. After adopting motion compensation for ASMS, the average grasping success rate is improved from 66.63% to 75.11%, and the stability of ASMS is increased by 6.35%. Furthermore, the average grasping time is decreased by 0.65 s, which is a 10.2% improvement in grasping efficiency.","PeriodicalId":13241,"journal":{"name":"IEEE Robotics and Automation Letters","volume":"10 6","pages":"5593-5600"},"PeriodicalIF":4.6,"publicationDate":"2025-04-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143888294","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

Mechanical Constraint Identification in Model-Mediated Teleoperation 模型介导遥操作中的机械约束识别

IF 4.6 2区计算机科学

IEEE Robotics and Automation Letters Pub Date : 2025-04-15 DOI: 10.1109/LRA.2025.3560893

Christophe van der Walt;Stefano Stramigioli;Douwe Dresscher

引用次数: 0

Global Grounding in Flexible Robot Tasks With Object Number Variety 具有目标数量变化的柔性机器人任务的全局接地

IF 4.6 2区计算机科学

IEEE Robotics and Automation Letters Pub Date : 2025-04-14 DOI: 10.1109/LRA.2025.3560864

Sascha Sucker;Dominik Henrich

{"title":"Global Grounding in Flexible Robot Tasks With Object Number Variety","authors":"Sascha Sucker;Dominik Henrich","doi":"10.1109/LRA.2025.3560864","DOIUrl":"https://doi.org/10.1109/LRA.2025.3560864","url":null,"abstract":"Automation in flexible environments must account for ambiguities and uncertainties. For example, the number of available objects may vary between different workspaces. To address this, we introduce flexible robot tasks with varieties that incorporate these ambiguities. This approach allows the programmer to estimate the number of required objects without precise knowledge of the world state during execution. With this, we deliberately leverage ambiguities, enabling task reuse across different world states. When executing a task with varieties, physical objects must be mapped to ambiguous object specifications called grounding. This grounding should be globally correct for the entire task and the world state. Rather than establishing a single grounding with fixed object numbers, we examine all the possible object numbers suitable for the task. Exhaustively testing every possibility would require exponential runtime. We overcome this challenge by contributing a global grounding algorithm for tasks with object number variety. Our algorithm uses the Kuhn-Munkres algorithm to establish fixed groundings and efficiently explores the problem space for flexible groundings – achieving a polynomial runtime. Through further optimization with binary search, our prototype demonstrates fast groundings (up to 252 objects in less than one second).","PeriodicalId":13241,"journal":{"name":"IEEE Robotics and Automation Letters","volume":"10 6","pages":"5879-5886"},"PeriodicalIF":4.6,"publicationDate":"2025-04-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10964876","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143901864","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

FreeDOM: Online Dynamic Object Removal Framework for Static Map Construction Based on Conservative Free Space Estimation FreeDOM：基于保守自由空间估计的静态地图构建在线动态目标移除框架

IF 4.6 2区计算机科学

IEEE Robotics and Automation Letters Pub Date : 2025-04-14 DOI: 10.1109/LRA.2025.3560881

Chen Li;Wanlei Li;Wenhao Liu;Yixiang Shu;Yunjiang Lou

{"title":"FreeDOM: Online Dynamic Object Removal Framework for Static Map Construction Based on Conservative Free Space Estimation","authors":"Chen Li;Wanlei Li;Wenhao Liu;Yixiang Shu;Yunjiang Lou","doi":"10.1109/LRA.2025.3560881","DOIUrl":"https://doi.org/10.1109/LRA.2025.3560881","url":null,"abstract":"Online map construction is essential for autonomous robots to navigate in unknown environments. However, the presence of dynamic objects may introduce artifacts into the map, which can significantly degrade the performance of localization and path planning. To tackle this problem, a novel online dynamic object removal framework for static map construction based on conservative free space estimation (FreeDOM) is proposed, consisting of a scan-removal front-end and a map-refinement back-end. First, we propose a multi-resolution map structure for fast computation and effective map representation. In the scan-removal front-end, we employ raycast enhancement to improve free space estimation and segment the LiDAR scan based on the estimated free space. In the map-refinement back-end, we further eliminate residual dynamic objects in the map by leveraging incremental free space information. As experimentally verified on SemanticKITTI, HeLiMOS, and indoor datasets with various sensors, our proposed framework overcomes the limitations of visibility-based methods and outperforms state-of-the-art methods with an average F1-score improvement of 9.7%.","PeriodicalId":13241,"journal":{"name":"IEEE Robotics and Automation Letters","volume":"10 6","pages":"5577-5584"},"PeriodicalIF":4.6,"publicationDate":"2025-04-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143888292","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

Temporal Logic-Guided DQN for Object Delivery and Transportation Using Magnetic Microrobots With Local Control 基于局部控制的磁性微机器人的时间逻辑引导DQN

IF 4.6 2区计算机科学

IEEE Robotics and Automation Letters Pub Date : 2025-04-14 DOI: 10.1109/LRA.2025.3560879

Zhe Hou;Yueyue Liu;Qigao Fan

{"title":"Temporal Logic-Guided DQN for Object Delivery and Transportation Using Magnetic Microrobots With Local Control","authors":"Zhe Hou;Yueyue Liu;Qigao Fan","doi":"10.1109/LRA.2025.3560879","DOIUrl":"https://doi.org/10.1109/LRA.2025.3560879","url":null,"abstract":"Microrobots driven by magnetic fields hold significant promise in medical applications, particularly in drug delivery–a key area in the biomedical field. Traditional approaches primarily rely on global magnetic fields to control single microrobots, which makes it challenging to independently control multiple microrobots. To address this limitation, this letter introduces a local magnetic field generation system that enables independent control of multiple microrobots. The proposed system employs a printed circuit board (PCB) array-based magnetic microrobot system, utilizing a microcoil array for precise and localized control. To enhance task execution, we integrate Deep Reinforcement Learning (DRL) with Linear Temporal Logic (LTL) to generate obstacle-avoiding paths for single and dual magnetic microrobots. The system is validated through magnetic droplet transport experiments. Experimental results demonstrate the effectiveness of the proposed system in achieving autonomous multi-task drug delivery and droplet fusion. This work underscores the potential of magnetic field-driven microcoil array systems in advancing transportation and drug delivery technologies in biomedical engineering.","PeriodicalId":13241,"journal":{"name":"IEEE Robotics and Automation Letters","volume":"10 6","pages":"5441-5448"},"PeriodicalIF":4.6,"publicationDate":"2025-04-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143888417","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 Predictor-Corrector Algorithm for the Fast Determination of the Wrench-Feasible Workspace of Cable-Driven Parallel Robots 一种快速确定缆索驱动并联机器人扳手可行工作空间的预测校正算法

IF 4.6 2区计算机科学

IEEE Robotics and Automation Letters Pub Date : 2025-04-14 DOI: 10.1109/LRA.2025.3560887

A. Abrous;P. Cardou

{"title":"A Predictor-Corrector Algorithm for the Fast Determination of the Wrench-Feasible Workspace of Cable-Driven Parallel Robots","authors":"A. Abrous;P. Cardou","doi":"10.1109/LRA.2025.3560887","DOIUrl":"https://doi.org/10.1109/LRA.2025.3560887","url":null,"abstract":"In this letter, we present a method of determining the Constant-Orientation Wrench-Feasible Workspace (COWFW) of Cable-Driven Parallel Robots (CDPRs). This workspace is a critical property of CDPRs, as their reach is often limited by their ability to sustain required forces and moments while keeping their cable tensions within acceptable limits. Several manners of determining the COWFW are reported in the literature, among which the brute force method is probably the most popular, the easiest to implement and the slowest, as it consists in determining wrench-feasibility at a grid of points covering the workspace. On the other hand, the ray-tracing method implemented in <monospace>WireX</monospace> software is probably the fastest method available; It consists in sampling the COWFW boundary by performing line-searches from a common starting point along predetermined, evenly-spaced directions. The method proposed in this letter is the application to CDPRs of a predictor-corrector algorithm used in computer graphics to quickly sample implicit surfaces In this work, the implicit surface is defined by equating to zero the “capacity margin” an index developed by one of the authors. The predictor-corrector algorithm then “marches” over the implicit surface, triangulating it as it advances. Simulation results show that the determination of the COWFW surface by this technique can be faster than the ray-tracing method in at least some examples, and often yields a triangulation that is more regular. We should note, however, that the implementation of this method is more demanding, it requiring more lines of code than the ray-tracing method.","PeriodicalId":13241,"journal":{"name":"IEEE Robotics and Automation Letters","volume":"10 6","pages":"5569-5576"},"PeriodicalIF":4.6,"publicationDate":"2025-04-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143888418","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