Pub Date : 2025-02-05DOI: 10.1109/LRA.2025.3539103
Nourhan Abdulazeem;Nils Sichert;Ji Yuan Feng;Yue Hu
As robots increasingly enter domestic environments, investigating the impact of their physical behaviors and the potential to leverage human mental states during interaction becomes crucial. This study examines how a robot's active behavior (unanticipated physical actions) versus passive behavior (actions aligned with user expectation) affects users' mental states during a physical balance task. Our findings show that passive interaction is generally more cognitively ergonomic, while active behavior, though it reduces imbalance, adds cognitive strain. Users' perceptions of the robot are not affected by its behavior type. We conclude that combining peripheral skin temperature with age and personality traits holds significant potential for enhancing robots' ability to infer users' cognitive ergonomics and belief levels. This study explores the relatively under-researched area of active behavior in physical assistive applications with minimal sensor requirements and identifies easily obtainable online data as indicators of human mental state.
{"title":"Quantifying Human Mental State in Interactive pHRI: Maintaining Balancing","authors":"Nourhan Abdulazeem;Nils Sichert;Ji Yuan Feng;Yue Hu","doi":"10.1109/LRA.2025.3539103","DOIUrl":"https://doi.org/10.1109/LRA.2025.3539103","url":null,"abstract":"As robots increasingly enter domestic environments, investigating the impact of their physical behaviors and the potential to leverage human mental states during interaction becomes crucial. This study examines how a robot's active behavior (unanticipated physical actions) versus passive behavior (actions aligned with user expectation) affects users' mental states during a physical balance task. Our findings show that passive interaction is generally more cognitively ergonomic, while active behavior, though it reduces imbalance, adds cognitive strain. Users' perceptions of the robot are not affected by its behavior type. We conclude that combining peripheral skin temperature with age and personality traits holds significant potential for enhancing robots' ability to infer users' cognitive ergonomics and belief levels. This study explores the relatively under-researched area of active behavior in physical assistive applications with minimal sensor requirements and identifies easily obtainable online data as indicators of human mental state.","PeriodicalId":13241,"journal":{"name":"IEEE Robotics and Automation Letters","volume":"10 3","pages":"2958-2965"},"PeriodicalIF":4.6,"publicationDate":"2025-02-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143430537","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 : 2025-02-05DOI: 10.1109/LRA.2025.3539118
Kavindie Katuwandeniya;Leimin Tian;Dana Kulić
This paper investigates the use of Video Foundation Models (ViFMs) for generating robot data summaries to enhance intermittent human supervision of robot teams. We propose a novel framework that produces both generic and query-driven summaries of long-duration robot vision data in three modalities: storyboards, short videos, and text. Through a user study involving 30 participants, we evaluate the efficacy of these summary methods in allowing operators to accurately retrieve the observations and actions that occurred while the robot was operating without supervision over an extended duration (40 min). Our findings reveal that query-driven summaries significantly improve retrieval accuracy compared to generic summaries or raw data, albeit with increased task duration. Storyboards are found to be the most effective presentation modality, especially for object-related queries.
{"title":"‘What Did the Robot Do in My Absence?’ Video Foundation Models to Enhance Intermittent Supervision","authors":"Kavindie Katuwandeniya;Leimin Tian;Dana Kulić","doi":"10.1109/LRA.2025.3539118","DOIUrl":"https://doi.org/10.1109/LRA.2025.3539118","url":null,"abstract":"This paper investigates the use of Video Foundation Models (ViFMs) for generating robot data summaries to enhance intermittent human supervision of robot teams. We propose a novel framework that produces both generic and query-driven summaries of long-duration robot vision data in three modalities: storyboards, short videos, and text. Through a user study involving 30 participants, we evaluate the efficacy of these summary methods in allowing operators to accurately retrieve the observations and actions that occurred while the robot was operating without supervision over an extended duration (40 min). Our findings reveal that query-driven summaries significantly improve retrieval accuracy compared to generic summaries or raw data, albeit with increased task duration. Storyboards are found to be the most effective presentation modality, especially for object-related queries.","PeriodicalId":13241,"journal":{"name":"IEEE Robotics and Automation Letters","volume":"10 4","pages":"3222-3229"},"PeriodicalIF":4.6,"publicationDate":"2025-02-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10873818","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143480855","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 : 2025-02-05DOI: 10.1109/LRA.2025.3539084
Yunjiang Wang;Keji Yang;Haoran Jin
Continuum and snake-like hyper-redundant robots perform circular bending motions, enable them to navigate obstacles and operate in confined spaces, making them ideal for applications such as industrial inspection and minimally invasive surgery. However, their performance often diminishes at the distal end, particularly in terms of positioning accuracy and output force during high-curvature tasks, which are considered their specialty. This letter introduces a novel hyper-redundant manipulator composed of zigzag-jointed folding links and intermediary connecting links. Each basic unit, termed a zigzag mechanism doublet (ZMD), consists of two folding links and their interacting connecting links, providing symmetric kinematic inputs and outputs. The output of one ZMD unit serves as the input for the next, enabling the entire manipulator to be actuated by the initial unit. By connecting multiple ZMD units, the manipulator approximates circular bending motion. This design outperforms traditional snake-like hyper-redundant manipulators in three aspects. First, it achieves bending motion through structural constraints, eliminating the need for tendons or other appendages to actuate multiple joints. Second, each unit extends the manipulator's motion range, rather than distributing a limited bending range across the entire structure. Third, the ZMD chain achieves constant curvature in discrete form, enhancing the manipulator's payload capability throughout its full motion range, even in extreme bending configurations. Experimental evaluations were conducted on a 3D-printed prototype and compared with typical articulated and continuum manipulators. The ZMD-based manipulator demonstrated a mean repeatability of 0.32 mm and a payload of 200 g, offering a promising solution for operations in constrained environments.
{"title":"Design of a Hyper-Redundant Manipulator With Zigzag Mechanism Doublet","authors":"Yunjiang Wang;Keji Yang;Haoran Jin","doi":"10.1109/LRA.2025.3539084","DOIUrl":"https://doi.org/10.1109/LRA.2025.3539084","url":null,"abstract":"Continuum and snake-like hyper-redundant robots perform circular bending motions, enable them to navigate obstacles and operate in confined spaces, making them ideal for applications such as industrial inspection and minimally invasive surgery. However, their performance often diminishes at the distal end, particularly in terms of positioning accuracy and output force during high-curvature tasks, which are considered their specialty. This letter introduces a novel hyper-redundant manipulator composed of zigzag-jointed folding links and intermediary connecting links. Each basic unit, termed a zigzag mechanism doublet (ZMD), consists of two folding links and their interacting connecting links, providing symmetric kinematic inputs and outputs. The output of one ZMD unit serves as the input for the next, enabling the entire manipulator to be actuated by the initial unit. By connecting multiple ZMD units, the manipulator approximates circular bending motion. This design outperforms traditional snake-like hyper-redundant manipulators in three aspects. First, it achieves bending motion through structural constraints, eliminating the need for tendons or other appendages to actuate multiple joints. Second, each unit extends the manipulator's motion range, rather than distributing a limited bending range across the entire structure. Third, the ZMD chain achieves constant curvature in discrete form, enhancing the manipulator's payload capability throughout its full motion range, even in extreme bending configurations. Experimental evaluations were conducted on a 3D-printed prototype and compared with typical articulated and continuum manipulators. The ZMD-based manipulator demonstrated a mean repeatability of 0.32 mm and a payload of 200 g, offering a promising solution for operations in constrained environments.","PeriodicalId":13241,"journal":{"name":"IEEE Robotics and Automation Letters","volume":"10 3","pages":"2990-2997"},"PeriodicalIF":4.6,"publicationDate":"2025-02-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143430506","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 : 2025-02-05DOI: 10.1109/LRA.2025.3539093
Yuhang Ming;Minyang Xu;Xingrui Yang;Weicai Ye;Weihan Wang;Yong Peng;Weichen Dai;Wanzeng Kong
Visual place recognition (VPR) is essential to many autonomous systems. Existing VPR methods demonstrate attractive performance at the cost of limited generalizability. When deployed in unseen environments, these methods exhibit significant performance drops. Targeting this issue, we present VIPeR, a novel approach for visual incremental place recognition with the ability to adapt to new environments while retaining the performance of previous ones. We first introduce an adaptive mining strategy that balances the performance within a single environment and the generalizability across multiple environments. Then, to prevent catastrophic forgetting in continual learning, we design a novel multi-stage memory bank for explicit rehearsal. Additionally, we propose a probabilistic knowledge distillation to explicitly safeguard the previously learned knowledge. We evaluate our proposed VIPeR on three large-scale datasets—Oxford Robotcar, Nordland, and TartanAir. For comparison, we first set a baseline performance with naive finetuning. Then, several more recent continual learning methods are compared. Our VIPeR achieves better performance in almost all aspects with the biggest improvement of 13.85% in average performance.
{"title":"VIPeR: Visual Incremental Place Recognition With Adaptive Mining and Continual Learning","authors":"Yuhang Ming;Minyang Xu;Xingrui Yang;Weicai Ye;Weihan Wang;Yong Peng;Weichen Dai;Wanzeng Kong","doi":"10.1109/LRA.2025.3539093","DOIUrl":"https://doi.org/10.1109/LRA.2025.3539093","url":null,"abstract":"Visual place recognition (VPR) is essential to many autonomous systems. Existing VPR methods demonstrate attractive performance at the cost of limited generalizability. When deployed in unseen environments, these methods exhibit significant performance drops. Targeting this issue, we present VIPeR, a novel approach for visual incremental place recognition with the ability to adapt to new environments while retaining the performance of previous ones. We first introduce an adaptive mining strategy that balances the performance within a single environment and the generalizability across multiple environments. Then, to prevent catastrophic forgetting in continual learning, we design a novel multi-stage memory bank for explicit rehearsal. Additionally, we propose a probabilistic knowledge distillation to explicitly safeguard the previously learned knowledge. We evaluate our proposed VIPeR on three large-scale datasets—Oxford Robotcar, Nordland, and TartanAir. For comparison, we first set a baseline performance with naive finetuning. Then, several more recent continual learning methods are compared. Our VIPeR achieves better performance in almost all aspects with the biggest improvement of 13.85% in average performance.","PeriodicalId":13241,"journal":{"name":"IEEE Robotics and Automation Letters","volume":"10 3","pages":"3038-3045"},"PeriodicalIF":4.6,"publicationDate":"2025-02-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143438346","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}
Compared with traditional in-pipe robots, tensegrity robots have exhibited many advantages such as light-weight, compliant, collapsible, low-cost, and rapidly manufacturable characteristics. However, published tensegrity in-pipe robots still have limited load capacity, because they rely on the stress between obliquely arranged rigid struts and pipe to provide supporting force, and the rigid struts are easy to slide under external loads. In this editor, a soft umbrella-shaped tensegrity robot that use compressed elastic struts to apply stress nearly perpendicular to pipe is proposed to enhance its load capacity. The static and kinematic models guiding the prototyping and controlling of the tensegrity robot are built based on discretization method. To assess the effectiveness of the elastic struts, a prototype is developed and subjected to a series of experiments. The results demonstrate that compared with other tensegrity in-pipe robots, the usage of elastic struts improves the payload-to-weight ratio of the proposed robot by three times, while still maintaining good mobility and adaptability.
{"title":"SUTBot: A Soft Umbrella-Like Tensegrity Robot With Elastic Struts for in-Pipe Locomotion","authors":"Yixiang Liu;Yunce Zhang;Xiaolin Dai;Rui Wu;Zhenyu Zhang;Yibin Li;Jie Zhao","doi":"10.1109/LRA.2025.3537866","DOIUrl":"https://doi.org/10.1109/LRA.2025.3537866","url":null,"abstract":"Compared with traditional in-pipe robots, tensegrity robots have exhibited many advantages such as light-weight, compliant, collapsible, low-cost, and rapidly manufacturable characteristics. However, published tensegrity in-pipe robots still have limited load capacity, because they rely on the stress between obliquely arranged rigid struts and pipe to provide supporting force, and the rigid struts are easy to slide under external loads. In this editor, a soft umbrella-shaped tensegrity robot that use compressed elastic struts to apply stress nearly perpendicular to pipe is proposed to enhance its load capacity. The static and kinematic models guiding the prototyping and controlling of the tensegrity robot are built based on discretization method. To assess the effectiveness of the elastic struts, a prototype is developed and subjected to a series of experiments. The results demonstrate that compared with other tensegrity in-pipe robots, the usage of elastic struts improves the payload-to-weight ratio of the proposed robot by three times, while still maintaining good mobility and adaptability.","PeriodicalId":13241,"journal":{"name":"IEEE Robotics and Automation Letters","volume":"10 3","pages":"2918-2925"},"PeriodicalIF":4.6,"publicationDate":"2025-02-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143430447","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}
The magnetic soft robot has potential applications in biomimetic, soft interaction, and biomedical fields. However, its functionality depends on deformation patterns and locomotion modes, challenging the fabrication of complex structures with precise magnetization. Therefore, we developed a programming magnetization photo-curing 3D printer for fabrication of small-scale soft robots. The printer integrates a three-dimensional magnetic field generator (3D-MFG) and a digital light processing photo-curing system. The 3D-MFG generates a high-strength (up to 80 mT) magnetic field through Halbach arrays (x-y plane) and a solenoid (z-axis), generating arbitrary uniform magnetic field with low energy consumption. Adhesion between the printed structure and the substrate was analyzed, and A real-time force-based printing control method is presented for precise optimization of key parameters, including layer thickness, approaching force, and separation speed, enhancing overall print quality and reliability in stacking of complex three-dimensional structures. Finally, a crawling robot mimicking inchworm gait, a swimming robot with butterfly-inspired motion, and a capsule robot for targeted drug delivery were fabricated by the developed system. These experimental results validated the printer's capability to fabricate highly complex structures, advancing the practical application of small-scale soft robots in biomimetic and biomedical fields.
{"title":"Development of a Photo-Curing 3D Printer for Fabrication of Small-Scale Soft Robots With Programming Spatial Magnetization","authors":"Shishi Li;Xianghe Meng;Xingjian Shen;Jinrong Wang;Hui Xie","doi":"10.1109/LRA.2025.3537857","DOIUrl":"https://doi.org/10.1109/LRA.2025.3537857","url":null,"abstract":"The magnetic soft robot has potential applications in biomimetic, soft interaction, and biomedical fields. However, its functionality depends on deformation patterns and locomotion modes, challenging the fabrication of complex structures with precise magnetization. Therefore, we developed a programming magnetization photo-curing 3D printer for fabrication of small-scale soft robots. The printer integrates a three-dimensional magnetic field generator (3D-MFG) and a digital light processing photo-curing system. The 3D-MFG generates a high-strength (up to 80 mT) magnetic field through Halbach arrays (<italic>x-y</i> plane) and a solenoid (<italic>z</i>-axis), generating arbitrary uniform magnetic field with low energy consumption. Adhesion between the printed structure and the substrate was analyzed, and A real-time force-based printing control method is presented for precise optimization of key parameters, including layer thickness, approaching force, and separation speed, enhancing overall print quality and reliability in stacking of complex three-dimensional structures. Finally, a crawling robot mimicking inchworm gait, a swimming robot with butterfly-inspired motion, and a capsule robot for targeted drug delivery were fabricated by the developed system. These experimental results validated the printer's capability to fabricate highly complex structures, advancing the practical application of small-scale soft robots in biomimetic and biomedical fields.","PeriodicalId":13241,"journal":{"name":"IEEE Robotics and Automation Letters","volume":"10 3","pages":"2766-2773"},"PeriodicalIF":4.6,"publicationDate":"2025-02-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143379605","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}
The inside-out patch clamp technique has been widely applied in brain science and neuroscience research due to its ability to detect extremely weak currents flowing through a single ion channel. The current manual inside-out patch clamp operations are highly expertise-requisite and low efficient. Meanwhile, the existing robotic systems are only applicable for suspended cells due to their new system setups. For the first time, this letter proposed a robotic inside-out patch clamp system for adherent cells based on vesicle rupture control. Firstly, impedance models were established to detect the vesicle rupture state. Then, a force analysis that combines the defocusing imaging model was developed to precisely control the exposure time of the vesicle in the air, which is a key factor in the rupture process of the vesicle. Based on the above works, a robotic inside-out patch clamp process for adherent cells was established. Experimental results demonstrate that the proposed robotic system can detect vesicle rupture state with a 100% success rate, control exposure time with an average error of 0.02$,text{s}$ and operate adherent HEK-293 cells with a success rate of 70% at an average operation speed of 61.3$,$seconds/cell. The success rate of our system is more than three times that of manual operation results, laying a solid foundation for subsequent single ion channel functionality research.
{"title":"Robotic Inside-Out Patch Clamp System for Adherent Cells Based on Vesicle Rupture Control","authors":"Yuzhu Liu;Ruimin Li;Jinyu Qiu;Biting Ma;Zuqi Wang;Minghui Li;Xin Zhao;Qili Zhao","doi":"10.1109/LRA.2025.3537870","DOIUrl":"https://doi.org/10.1109/LRA.2025.3537870","url":null,"abstract":"The inside-out patch clamp technique has been widely applied in brain science and neuroscience research due to its ability to detect extremely weak currents flowing through a single ion channel. The current manual inside-out patch clamp operations are highly expertise-requisite and low efficient. Meanwhile, the existing robotic systems are only applicable for suspended cells due to their new system setups. For the first time, this letter proposed a robotic inside-out patch clamp system for adherent cells based on vesicle rupture control. Firstly, impedance models were established to detect the vesicle rupture state. Then, a force analysis that combines the defocusing imaging model was developed to precisely control the exposure time of the vesicle in the air, which is a key factor in the rupture process of the vesicle. Based on the above works, a robotic inside-out patch clamp process for adherent cells was established. Experimental results demonstrate that the proposed robotic system can detect vesicle rupture state with a 100% success rate, control exposure time with an average error of 0.02<inline-formula><tex-math>$,text{s}$</tex-math></inline-formula> and operate adherent HEK-293 cells with a success rate of 70% at an average operation speed of 61.3<inline-formula><tex-math>$,$</tex-math></inline-formula>seconds/cell. The success rate of our system is more than three times that of manual operation results, laying a solid foundation for subsequent single ion channel functionality research.","PeriodicalId":13241,"journal":{"name":"IEEE Robotics and Automation Letters","volume":"10 3","pages":"3014-3021"},"PeriodicalIF":4.6,"publicationDate":"2025-02-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143438512","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 : 2025-02-03DOI: 10.1109/LRA.2025.3537899
Yang You;Bokui Shen;Congyue Deng;Haoran Geng;Songlin Wei;He Wang;Leonidas Guibas
Deformable object manipulation stands as one of the most captivating yet formidable challenges in robotics. While previous techniques have predominantly relied on learning latent dynamics through demonstrations, typically represented as either particles or images, there exists a pertinent limitation: acquiring suitable demonstrations, especially for long-horizon tasks, can be elusive. Moreover, basing learning entirely on demonstrations can hamper the model's ability to generalize beyond the demonstrated tasks. In this work, we introduce a demonstration-free hierarchical planning approach capable of tackling intricate long-horizon deformable manipulation tasks without necessitating any training. We employ large language models (LLMs) to articulate a high-level, stage-by-stage plan corresponding to a specified task. For every individual stage, the LLM provides both the tool's name and the Python code to craft intermediate subgoal point clouds. With the tool and subgoal for a particular stage at our disposal, we present a granular closed-loop model predictive control strategy. This leverages Differentiable Physics with Point-to-Point correspondence (DiffPhysics-P2P) loss in the Earth Mover Distance (EMD) space, applied iteratively. Experimental findings affirm that our technique surpasses multiple benchmarks in dough manipulation, spanning both short and long horizons. Remarkably, our model demonstrates robust generalization capabilities to novel and previously unencountered complex tasks without any preliminary demonstrations. We further substantiate our approach with experimental trials on real-world robotic platforms.
{"title":"Make a Donut: Hierarchical EMD-Space Planning for Zero-Shot Deformable Manipulation With Tools","authors":"Yang You;Bokui Shen;Congyue Deng;Haoran Geng;Songlin Wei;He Wang;Leonidas Guibas","doi":"10.1109/LRA.2025.3537899","DOIUrl":"https://doi.org/10.1109/LRA.2025.3537899","url":null,"abstract":"Deformable object manipulation stands as one of the most captivating yet formidable challenges in robotics. While previous techniques have predominantly relied on learning latent dynamics through demonstrations, typically represented as either particles or images, there exists a pertinent limitation: acquiring suitable demonstrations, especially for long-horizon tasks, can be elusive. Moreover, basing learning entirely on demonstrations can hamper the model's ability to generalize beyond the demonstrated tasks. In this work, we introduce a demonstration-free hierarchical planning approach capable of tackling intricate long-horizon deformable manipulation tasks without necessitating any training. We employ large language models (LLMs) to articulate a high-level, stage-by-stage plan corresponding to a specified task. For every individual stage, the LLM provides both the tool's name and the Python code to craft intermediate subgoal point clouds. With the tool and subgoal for a particular stage at our disposal, we present a granular closed-loop model predictive control strategy. This leverages Differentiable Physics with Point-to-Point correspondence (DiffPhysics-P2P) loss in the Earth Mover Distance (EMD) space, applied iteratively. Experimental findings affirm that our technique surpasses multiple benchmarks in dough manipulation, spanning both short and long horizons. Remarkably, our model demonstrates robust generalization capabilities to novel and previously unencountered complex tasks without any preliminary demonstrations. We further substantiate our approach with experimental trials on real-world robotic platforms.","PeriodicalId":13241,"journal":{"name":"IEEE Robotics and Automation Letters","volume":"10 4","pages":"3270-3277"},"PeriodicalIF":4.6,"publicationDate":"2025-02-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143489206","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 : 2025-02-03DOI: 10.1109/LRA.2025.3537871
Jun Zhang;Haoyun Chen;Yun Li;Xiaoyu Chu;Xiaodong Zhou;Aiguo Song
Continuum robots (CRs) possess better compliance than rigid manipulators. However, existing CRs suffer from difficulties in manipulating objects for the conflicting needs of high stiffness and flexibility. This letter proposes an elephant trunk-inspired CR for grasping and moving objects. The CR features a deformation mechanism based on coupling rigid discs and flexible elastic rods and a stiffness modulation mechanism (SMM) with a force amplification function. The SMM decouples the CR's stiffness adjustment from bending motion, achieving flexible control capabilities. The SMM uses a stiffness coding strategy (SCS) to adjust stiffness according to task requirements to improve grasp performance while maintaining flexibility. We established the robot's kinematics, friction, and coiling models, designed a control method to coil around and move objects, and simulated the robot's end position-pose space and object size-pose space. We designed a robot prototype and conducted various experiments. Results showed that the robot achieved free movement in randomly generated configurations, coiled around and moved various objects with a path deviation of less than 8.5%, increased stiffness to 206%, and increased grasp force by 31%, demonstrating its potential in real applications.
{"title":"Design of a Bio-Inspired Stiffness Controllable Continuum Robot for Object Grasping and Moving","authors":"Jun Zhang;Haoyun Chen;Yun Li;Xiaoyu Chu;Xiaodong Zhou;Aiguo Song","doi":"10.1109/LRA.2025.3537871","DOIUrl":"https://doi.org/10.1109/LRA.2025.3537871","url":null,"abstract":"Continuum robots (CRs) possess better compliance than rigid manipulators. However, existing CRs suffer from difficulties in manipulating objects for the conflicting needs of high stiffness and flexibility. This letter proposes an elephant trunk-inspired CR for grasping and moving objects. The CR features a deformation mechanism based on coupling rigid discs and flexible elastic rods and a stiffness modulation mechanism (SMM) with a force amplification function. The SMM decouples the CR's stiffness adjustment from bending motion, achieving flexible control capabilities. The SMM uses a stiffness coding strategy (SCS) to adjust stiffness according to task requirements to improve grasp performance while maintaining flexibility. We established the robot's kinematics, friction, and coiling models, designed a control method to coil around and move objects, and simulated the robot's end position-pose space and object size-pose space. We designed a robot prototype and conducted various experiments. Results showed that the robot achieved free movement in randomly generated configurations, coiled around and moved various objects with a path deviation of less than 8.5%, increased stiffness to 206%, and increased grasp force by 31%, demonstrating its potential in real applications.","PeriodicalId":13241,"journal":{"name":"IEEE Robotics and Automation Letters","volume":"10 3","pages":"2982-2989"},"PeriodicalIF":4.6,"publicationDate":"2025-02-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143430575","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 : 2025-01-31DOI: 10.1109/LRA.2025.3537860
Sahar Leisiazar;Seyed Roozbeh Razavi Rohani;Edward J. Park;Angelica Lim;Mo Chen
This letter addresses the challenge of robot follow ahead applications where the human behavior is highly variable. We propose a novel approach that does not rely on single human trajectory prediction but instead considers multiple potential future positions of the human, along with their associated probabilities, in the robot's decision-making process. We trained an LSTM-based model to generate a probability distribution over the human's future actions. These probabilities, along with different potential actions and future positions, are integrated into the tree expansion of Monte Carlo Tree Search (MCTS). Additionally, a trained Reinforcement Learning (RL) model is used to evaluate the nodes within the tree. By incorporating the likelihood of each possible human action and using the RL model to assess the value of the different trajectories, our approach enables the robot to effectively balance between focusing on the most probable future trajectory and considering all potential trajectories. This methodology enhances the robot's ability to adapt to frequent and unpredictable changes in human direction, improving its navigation and ability to navigate in front of the person.
{"title":"Adapting to Frequent Human Direction Changes in Autonomous Frontal Following Robots","authors":"Sahar Leisiazar;Seyed Roozbeh Razavi Rohani;Edward J. Park;Angelica Lim;Mo Chen","doi":"10.1109/LRA.2025.3537860","DOIUrl":"https://doi.org/10.1109/LRA.2025.3537860","url":null,"abstract":"This letter addresses the challenge of robot follow ahead applications where the human behavior is highly variable. We propose a novel approach that does not rely on single human trajectory prediction but instead considers multiple potential future positions of the human, along with their associated probabilities, in the robot's decision-making process. We trained an LSTM-based model to generate a probability distribution over the human's future actions. These probabilities, along with different potential actions and future positions, are integrated into the tree expansion of Monte Carlo Tree Search (MCTS). Additionally, a trained Reinforcement Learning (RL) model is used to evaluate the nodes within the tree. By incorporating the likelihood of each possible human action and using the RL model to assess the value of the different trajectories, our approach enables the robot to effectively balance between focusing on the most probable future trajectory and considering all potential trajectories. This methodology enhances the robot's ability to adapt to frequent and unpredictable changes in human direction, improving its navigation and ability to navigate in front of the person.","PeriodicalId":13241,"journal":{"name":"IEEE Robotics and Automation Letters","volume":"10 3","pages":"2934-2941"},"PeriodicalIF":4.6,"publicationDate":"2025-01-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143430570","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：