Pub Date : 2025-12-26DOI: 10.1109/LRA.2025.3648610
Shuhuan Wen;Songhao Tan;Xin Liu;Mengyu Li;Huaping Liu
Visual simultaneous localization and mapping (VSLAM) is a foundational technology in robotics, providing an optimal balance of cost and accuracy. However, existing systems often lack robustness in environments with fast motion, dynamic lighting, or low texture. This letter introduces ML-SLAM, a hybrid visual-inertial SLAM system that combines point-line features with learning-based techniques to improve performance in these challenging conditions. Built on the ORB-SLAM3 framework, ML-SLAM incorporates SuperPoint for adaptive keypoint detection and LightGlue for robust feature matching, along with a novel endpoint-based point-line association strategy to enhance tracking reliability in complex scenes. The system also features hybrid feature-based loop-closure detection and tightly coupled bundle adjustment (BA) incorporating inertial measurements, adapted as standard modules in the ORB-SLAM3 backend to seamlessly integrate the hybrid point-line frontend with the established backend. Extensive evaluations on the EuRoC, TartanAir, UMA-VI, and real-world indoor datasets show that ML-SLAM significantly outperforms state-of-the-art (SOTA) methods, with over 20% improvement in localization accuracy compared to ORB-SLAM3.
{"title":"A Robust and Efficient Visual-Inertial SLAM Using Hybrid Point-Line Features","authors":"Shuhuan Wen;Songhao Tan;Xin Liu;Mengyu Li;Huaping Liu","doi":"10.1109/LRA.2025.3648610","DOIUrl":"https://doi.org/10.1109/LRA.2025.3648610","url":null,"abstract":"Visual simultaneous localization and mapping (VSLAM) is a foundational technology in robotics, providing an optimal balance of cost and accuracy. However, existing systems often lack robustness in environments with fast motion, dynamic lighting, or low texture. This letter introduces ML-SLAM, a hybrid visual-inertial SLAM system that combines point-line features with learning-based techniques to improve performance in these challenging conditions. Built on the ORB-SLAM3 framework, ML-SLAM incorporates SuperPoint for adaptive keypoint detection and LightGlue for robust feature matching, along with a novel endpoint-based point-line association strategy to enhance tracking reliability in complex scenes. The system also features hybrid feature-based loop-closure detection and tightly coupled bundle adjustment (BA) incorporating inertial measurements, adapted as standard modules in the ORB-SLAM3 backend to seamlessly integrate the hybrid point-line frontend with the established backend. Extensive evaluations on the EuRoC, TartanAir, UMA-VI, and real-world indoor datasets show that ML-SLAM significantly outperforms state-of-the-art (SOTA) methods, with over 20% improvement in localization accuracy compared to ORB-SLAM3.","PeriodicalId":13241,"journal":{"name":"IEEE Robotics and Automation Letters","volume":"11 2","pages":"2258-2265"},"PeriodicalIF":5.3,"publicationDate":"2025-12-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145929630","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}
Pub Date : 2025-12-26DOI: 10.1109/LRA.2025.3648502
Ziwon Yoon;Lawrence Y. Zhu;Jingxi Lu;Lu Gan;Ye Zhao
Bipedal robots have advantages in maneuvering human-centered environments, but face greater failure risk compared to other stable mobile platforms, such as wheeled or quadrupedal robots. While learning-based traversability has been widely studied for these platforms, bipedal traversability has instead relied on manually designed rules with limited consideration of locomotion stability on rough terrain. In this work, we present the first learning-based traversability estimation and risk-sensitive navigation framework for bipedal robots operating in diverse, uneven environments. TravFormer, a transformer-based neural network, is trained to predict bipedal instability with uncertainty, enabling risk-aware and adaptive planning. Based on the network, we define traversability as stability-aware command velocity—the fastest command velocity that keeps instability below a user-defined limit. This velocity-based traversability is integrated into a hierarchical planner that combines traversability-informed Rapid Random Tree Star (TravRRT*) for time-efficient path planning and Model Predictive Control (MPC) for safe execution. We validate our method in MuJoCo simulator and the real world, demonstrating improved stability, time efficiency, and robustness across diverse terrains compared with existing methods.
{"title":"STATE-NAV: Stability-Aware Traversability Estimation for Bipedal Navigation on Rough Terrain","authors":"Ziwon Yoon;Lawrence Y. Zhu;Jingxi Lu;Lu Gan;Ye Zhao","doi":"10.1109/LRA.2025.3648502","DOIUrl":"https://doi.org/10.1109/LRA.2025.3648502","url":null,"abstract":"Bipedal robots have advantages in maneuvering human-centered environments, but face greater failure risk compared to other stable mobile platforms, such as wheeled or quadrupedal robots. While learning-based traversability has been widely studied for these platforms, bipedal traversability has instead relied on manually designed rules with limited consideration of locomotion stability on rough terrain. In this work, we present the first learning-based traversability estimation and risk-sensitive navigation framework for bipedal robots operating in diverse, uneven environments. TravFormer, a transformer-based neural network, is trained to predict bipedal instability with uncertainty, enabling risk-aware and adaptive planning. Based on the network, we define traversability as stability-aware command velocity—the fastest command velocity that keeps instability below a user-defined limit. This velocity-based traversability is integrated into a hierarchical planner that combines traversability-informed Rapid Random Tree Star (TravRRT*) for time-efficient path planning and Model Predictive Control (MPC) for safe execution. We validate our method in MuJoCo simulator and the real world, demonstrating improved stability, time efficiency, and robustness across diverse terrains compared with existing methods.","PeriodicalId":13241,"journal":{"name":"IEEE Robotics and Automation Letters","volume":"11 2","pages":"2338-2345"},"PeriodicalIF":5.3,"publicationDate":"2025-12-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146026361","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}
Pub Date : 2025-12-25DOI: 10.1109/LRA.2025.3648605
M. Faroni;A. Spanò;A. M. Zanchettin;P. Rocco
Ensuring human safety in collaborative robotics can compromise efficiency because traditional safety measures increase robot cycle time when human interaction is frequent. This letter proposes a safety-aware approach to mitigate efficiency losses without assuming prior knowledge of safety logic. Using a deep-learning model, the robot learns the relationship between system state and safety-induced speed reductions based on execution data. Our framework does not explicitly predict human motions but directly models the interaction effects on robot speed, simplifying implementation and enhancing generalizability to different safety logics. At runtime, the learned model optimizes task selection to minimize cycle time while adhering to safety requirements. Experiments on a pick-and-packaging scenario demonstrated significant reductions in cycle times.
{"title":"Learning-Based Safety-Aware Task Scheduling for Efficient Human-Robot Collaboration","authors":"M. Faroni;A. Spanò;A. M. Zanchettin;P. Rocco","doi":"10.1109/LRA.2025.3648605","DOIUrl":"https://doi.org/10.1109/LRA.2025.3648605","url":null,"abstract":"Ensuring human safety in collaborative robotics can compromise efficiency because traditional safety measures increase robot cycle time when human interaction is frequent. This letter proposes a safety-aware approach to mitigate efficiency losses without assuming prior knowledge of safety logic. Using a deep-learning model, the robot learns the relationship between system state and safety-induced speed reductions based on execution data. Our framework does not explicitly predict human motions but directly models the interaction effects on robot speed, simplifying implementation and enhancing generalizability to different safety logics. At runtime, the learned model optimizes task selection to minimize cycle time while adhering to safety requirements. Experiments on a pick-and-packaging scenario demonstrated significant reductions in cycle times.","PeriodicalId":13241,"journal":{"name":"IEEE Robotics and Automation Letters","volume":"11 2","pages":"2226-2233"},"PeriodicalIF":5.3,"publicationDate":"2025-12-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146082198","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}
Pub Date : 2025-12-25DOI: 10.1109/LRA.2025.3648503
Shiming He;Yuzhe Ding
We propose a novel framework for data-efficient black-box robot learning under constraints. Our approach integrates probabilistic inference with Lagrangian optimization. With the guide of a learned Gaussian process model, the Lagrange multiplier is controlled by the probability of whether the constraints would be satisfied. This reduces the typical oscillations seen in primal-dual updates and therefore improves both data efficiency and safety during learning. Both synthetic results and robot experiments demonstrate that our method is a scalable and effective solution for constrained robot learning problems.
{"title":"Data-Efficient Constrained Robot Learning With Probabilistic Lagrangian Control","authors":"Shiming He;Yuzhe Ding","doi":"10.1109/LRA.2025.3648503","DOIUrl":"https://doi.org/10.1109/LRA.2025.3648503","url":null,"abstract":"We propose a novel framework for data-efficient black-box robot learning under constraints. Our approach integrates probabilistic inference with Lagrangian optimization. With the guide of a learned Gaussian process model, the Lagrange multiplier is controlled by the probability of whether the constraints would be satisfied. This reduces the typical oscillations seen in primal-dual updates and therefore improves both data efficiency and safety during learning. Both synthetic results and robot experiments demonstrate that our method is a scalable and effective solution for constrained robot learning problems.","PeriodicalId":13241,"journal":{"name":"IEEE Robotics and Automation Letters","volume":"11 2","pages":"2154-2161"},"PeriodicalIF":5.3,"publicationDate":"2025-12-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145886647","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}
Pub Date : 2025-12-25DOI: 10.1109/LRA.2025.3648604
Emanuele Solfiti;Alessio Mondini;Emanuela Del Dottore;Barbara Mazzolai;Alberto Parmiggiani
This paper presents the design, development, and testing of a soft 3D-printed endoskeleton for arbitrary cable routing in tendon-driven soft actuators. The endoskeleton is embedded in a silicone body, and it is fixed to the mold prior to the casting process. It enables tendons to be placed through predefined eyelets, ensuring accurate positioning within the soft body. To minimize its impact on the overall stiffness of the soft body, the endoskeleton was designed with a slim profile, flexible connections, and 3D-printed with elastic material (Shore A hardness 50), selected to roughly match the mechanical properties of the surrounding silicone matrix (typically with Shore 00 hardness 20–30). Key features of the proposed solution include a 3D-printable guide for tendon routing that is (1) fully soft, (2) easy to place, (3) rapidly reconfigurable for arbitrary tendon paths, (4) adaptable to variable soft body geometries, and (5) easy to fabricate with single-step casting. The current work describes the design, manufacturing, simulation, and testing of a simplified case study in which the endoskeleton is employed to reproduce a target pose predicted by FE analysis with good matching, demonstrating the effectiveness of the approach.
{"title":"Soft 3D-Printed Endoskeleton for Precise Tendon Routing in Soft Robotics","authors":"Emanuele Solfiti;Alessio Mondini;Emanuela Del Dottore;Barbara Mazzolai;Alberto Parmiggiani","doi":"10.1109/LRA.2025.3648604","DOIUrl":"https://doi.org/10.1109/LRA.2025.3648604","url":null,"abstract":"This paper presents the design, development, and testing of a soft 3D-printed endoskeleton for arbitrary cable routing in tendon-driven soft actuators. The endoskeleton is embedded in a silicone body, and it is fixed to the mold prior to the casting process. It enables tendons to be placed through predefined eyelets, ensuring accurate positioning within the soft body. To minimize its impact on the overall stiffness of the soft body, the endoskeleton was designed with a slim profile, flexible connections, and 3D-printed with elastic material (Shore A hardness 50), selected to roughly match the mechanical properties of the surrounding silicone matrix (typically with Shore 00 hardness 20–30). Key features of the proposed solution include a 3D-printable guide for tendon routing that is (1) fully soft, (2) easy to place, (3) rapidly reconfigurable for arbitrary tendon paths, (4) adaptable to variable soft body geometries, and (5) easy to fabricate with single-step casting. The current work describes the design, manufacturing, simulation, and testing of a simplified case study in which the endoskeleton is employed to reproduce a target pose predicted by FE analysis with good matching, demonstrating the effectiveness of the approach.","PeriodicalId":13241,"journal":{"name":"IEEE Robotics and Automation Letters","volume":"11 2","pages":"2282-2289"},"PeriodicalIF":5.3,"publicationDate":"2025-12-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11315156","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145982292","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}
Pub Date : 2025-12-19DOI: 10.1109/LRA.2025.3642593
{"title":"IEEE Robotics and Automation Society Information","authors":"","doi":"10.1109/LRA.2025.3642593","DOIUrl":"https://doi.org/10.1109/LRA.2025.3642593","url":null,"abstract":"","PeriodicalId":13241,"journal":{"name":"IEEE Robotics and Automation Letters","volume":"11 1","pages":"C3-C3"},"PeriodicalIF":5.3,"publicationDate":"2025-12-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11306189","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145778450","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}
Pub Date : 2025-12-19DOI: 10.1109/LRA.2025.3642595
{"title":"IEEE Robotics and Automation Letters Information for Authors","authors":"","doi":"10.1109/LRA.2025.3642595","DOIUrl":"https://doi.org/10.1109/LRA.2025.3642595","url":null,"abstract":"","PeriodicalId":13241,"journal":{"name":"IEEE Robotics and Automation Letters","volume":"11 1","pages":"C4-C4"},"PeriodicalIF":5.3,"publicationDate":"2025-12-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11306223","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145778452","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}
Pub Date : 2025-12-18DOI: 10.1109/LRA.2025.3645685
Donipolo Ghimire;Aamodh Suresh;Carlos Nieto-Granda;Solmaz S. Kia
This letter presents BEASST (Behavioral Entropic Gradient-based Adaptive Source Seeking for Mobile Robots), a novel framework for robotic source seeking in complex, unknown environments. Our approach enables mobile robots to efficiently balance exploration and exploitation by modeling normalized signal strength as a surrogate probability of source location. Building on Behavioral Entropy (BE) with Prelec's probability weighting function, we define an objective function that adapts robot behavior from risk-averse to risk-seeking based on signal reliability and mission urgency. The framework provides theoretical convergence guarantees under unimodal signal assumptions and practical stability under bounded disturbances. Experimental validation across DARPA SubT and multi-room scenarios demonstrates that BEASST consistently outperforms state-of-the-art methods and exhibits strong robustness to noisy gradient estimates while maintaining convergence. BEASST achieved 15% reduction in path length and 20% faster source localization through intelligent uncertainty-driven navigation that dynamically transitions between aggressive pursuit and cautious exploration.
{"title":"BEASST: Behavioral Entropic Gradient Based Adaptive Source Seeking for Mobile Robots","authors":"Donipolo Ghimire;Aamodh Suresh;Carlos Nieto-Granda;Solmaz S. Kia","doi":"10.1109/LRA.2025.3645685","DOIUrl":"https://doi.org/10.1109/LRA.2025.3645685","url":null,"abstract":"This letter presents BEASST (Behavioral Entropic Gradient-based Adaptive Source Seeking for Mobile Robots), a novel framework for robotic source seeking in complex, unknown environments. Our approach enables mobile robots to efficiently balance exploration and exploitation by modeling normalized signal strength as a surrogate probability of source location. Building on Behavioral Entropy (BE) with Prelec's probability weighting function, we define an objective function that adapts robot behavior from risk-averse to risk-seeking based on signal reliability and mission urgency. The framework provides theoretical convergence guarantees under unimodal signal assumptions and practical stability under bounded disturbances. Experimental validation across DARPA SubT and multi-room scenarios demonstrates that BEASST consistently outperforms state-of-the-art methods and exhibits strong robustness to noisy gradient estimates while maintaining convergence. BEASST achieved 15% reduction in path length and 20% faster source localization through intelligent uncertainty-driven navigation that dynamically transitions between aggressive pursuit and cautious exploration.","PeriodicalId":13241,"journal":{"name":"IEEE Robotics and Automation Letters","volume":"11 2","pages":"1906-1913"},"PeriodicalIF":5.3,"publicationDate":"2025-12-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145830770","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}
Pub Date : 2025-12-18DOI: 10.1109/LRA.2025.3645990
Achala Athukorala;Billy Pik Lik Lau;Khattiya Pongsirijinda;Chau Yuen;U-Xuan Tan
Autonomous mobile robot systems have been extremely useful in exploration tasks for inspection and surveying of unknown environments, where map quality and exploration speed are often important factors. To effectively increase the exploration speed, multi-robot systems and collaborative exploration have been gaining attention in recent years. However, multi-robot exploration introduces two main challenges: 1) shared mapping between the robots; and 2) efficient coordination between the robots. Towards efficient and practical multi-robot exploration, this work proposes a new Distributed Multi-Robot Collaborative SLAM (Multi-SLAM) framework and a Lightweight Predictive Frontier Exploration (LPFE) to enable ground robot fleets to explore unknown environments faster and efficiently. Our Multi-SLAM approach generates a graph based globally optimized map using information from all robots in the environment in a network bandwidth efficient manner, while our LPFE coordinates the exploration of the robots using a deterministic, inference-based heuristic, allowing robots to anticipate one another's actions without explicit communication. The experimental results demonstrate that our pipeline outperforms traditional frontier exploration approach, as well as state-of-the-art planners for ground robots, with up to 70% reduction in exploration times, with up to $13times$ less CPU usage and up to $50times$ less network bandwidth usage. We also present our Multi-SLAM and LPFE code-base which we have extensively tested in real-world robot fleets in different environments.
{"title":"Multi-Robot Collaborative SLAM (Multi-SLAM) With Distributed Lightweight Predictive Frontier Exploration (LPFE)","authors":"Achala Athukorala;Billy Pik Lik Lau;Khattiya Pongsirijinda;Chau Yuen;U-Xuan Tan","doi":"10.1109/LRA.2025.3645990","DOIUrl":"https://doi.org/10.1109/LRA.2025.3645990","url":null,"abstract":"Autonomous mobile robot systems have been extremely useful in exploration tasks for inspection and surveying of unknown environments, where map quality and exploration speed are often important factors. To effectively increase the exploration speed, multi-robot systems and collaborative exploration have been gaining attention in recent years. However, multi-robot exploration introduces two main challenges: 1) shared mapping between the robots; and 2) efficient coordination between the robots. Towards efficient and practical multi-robot exploration, this work proposes a new Distributed Multi-Robot Collaborative SLAM (Multi-SLAM) framework and a Lightweight Predictive Frontier Exploration (LPFE) to enable ground robot fleets to explore unknown environments faster and efficiently. Our Multi-SLAM approach generates a graph based globally optimized map using information from all robots in the environment in a network bandwidth efficient manner, while our LPFE coordinates the exploration of the robots using a deterministic, inference-based heuristic, allowing robots to anticipate one another's actions without explicit communication. The experimental results demonstrate that our pipeline outperforms traditional frontier exploration approach, as well as state-of-the-art planners for ground robots, with up to 70% reduction in exploration times, with up to <inline-formula><tex-math>$13times$</tex-math></inline-formula> less CPU usage and up to <inline-formula><tex-math>$50times$</tex-math></inline-formula> less network bandwidth usage. We also present our Multi-SLAM and LPFE code-base which we have extensively tested in real-world robot fleets in different environments.","PeriodicalId":13241,"journal":{"name":"IEEE Robotics and Automation Letters","volume":"11 2","pages":"2274-2281"},"PeriodicalIF":5.3,"publicationDate":"2025-12-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145929631","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}
Pub Date : 2025-12-18DOI: 10.1109/LRA.2025.3645668
Zichen Yan;Rui Huang;Lei He;Shao Guo;Lin Zhao
Image-goal navigation (ImageNav) tasks a robot with autonomously exploring an unknown environment and reaching a location that visually matches a given target image. While prior works primarily study ImageNav for ground robots, enabling this capability for autonomous drones is substantially more challenging due to their need for high-frequency feedback control and global localization for stable flight. In this letter, we propose a novel sim-to-real framework that leverages reinforcement learning (RL) to achieve ImageNav for drones. To enhance visual representation ability, our approach trains the vision backbone with auxiliary tasks, including image perturbations and future transition prediction, which results in more effective policy training. The proposed algorithm enables end-to-end ImageNav with direct velocity control, eliminating the need for external localization. Furthermore, we integrate a depth-based safety module for real-time obstacle avoidance, allowing the drone to safely navigate in cluttered environments. Unlike most existing drone navigation methods that focus solely on reference tracking or obstacle avoidance, our framework supports comprehensive navigation behaviors, including autonomous exploration, obstacle avoidance, and image-goal seeking, without requiring explicit global mapping.
{"title":"SIGN: Safety-Aware Image-Goal Navigation for Autonomous Drones via Reinforcement Learning","authors":"Zichen Yan;Rui Huang;Lei He;Shao Guo;Lin Zhao","doi":"10.1109/LRA.2025.3645668","DOIUrl":"https://doi.org/10.1109/LRA.2025.3645668","url":null,"abstract":"Image-goal navigation (ImageNav) tasks a robot with autonomously exploring an unknown environment and reaching a location that visually matches a given target image. While prior works primarily study ImageNav for ground robots, enabling this capability for autonomous drones is substantially more challenging due to their need for high-frequency feedback control and global localization for stable flight. In this letter, we propose a novel sim-to-real framework that leverages reinforcement learning (RL) to achieve ImageNav for drones. To enhance visual representation ability, our approach trains the vision backbone with auxiliary tasks, including image perturbations and future transition prediction, which results in more effective policy training. The proposed algorithm enables end-to-end ImageNav with direct velocity control, eliminating the need for external localization. Furthermore, we integrate a depth-based safety module for real-time obstacle avoidance, allowing the drone to safely navigate in cluttered environments. Unlike most existing drone navigation methods that focus solely on reference tracking or obstacle avoidance, our framework supports comprehensive navigation behaviors, including autonomous exploration, obstacle avoidance, and image-goal seeking, without requiring explicit global mapping.","PeriodicalId":13241,"journal":{"name":"IEEE Robotics and Automation Letters","volume":"11 2","pages":"1962-1969"},"PeriodicalIF":5.3,"publicationDate":"2025-12-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145830921","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}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：