Pub Date : 2024-12-23DOI: 10.1109/LRA.2024.3521180
Shuo Liu;Shihao Shen;Yunshan Li;Ming Xu
Soft pneumatic joints are the main driving components of soft arthropod robots, but the problems of high start-up air pressure and low rotation angle limit the range of motion, control accuracy, and flexibility of the robots. Inspired by the hydraulic joints of spider leg, an innovative folding membrane soft joint actuator is designed to achieve an 80° rotation angle at 5.25 kPa air pressure, with an angle density of 15.23°/ kPa and a fatigue life of more than 10,000 times, significantly improving the output characteristics of pneumatic joints. This actuator's folding membrane is made of silicone-coated fiber fabric and is manufactured via integrated molding method. Furthermore, regarding the issue of ineffective expansion in actuators, the effects of various constraint layer materials on the actuator's output performance are investigated and discussed. It is found that the TPU-constrained actuator has a maximum output torque of 0.307 N·m, which is 7.7% and 13.7% greater than that of the fiber-constrained and unconstrained actuators, respectively. Meanwhile, the rebound time was 1.25 seconds, which was reduced by 0.23 seconds and 0.22 seconds respectively. Soft actuators can benefit from the design technique of the folding membrane soft joint actuator presented in the study, which can provide significant torque output at low air pressure and has superior fatigue characteristics.
{"title":"Spider-Inspired Pneumatic Folding Membrane Soft Actuator","authors":"Shuo Liu;Shihao Shen;Yunshan Li;Ming Xu","doi":"10.1109/LRA.2024.3521180","DOIUrl":"https://doi.org/10.1109/LRA.2024.3521180","url":null,"abstract":"Soft pneumatic joints are the main driving components of soft arthropod robots, but the problems of high start-up air pressure and low rotation angle limit the range of motion, control accuracy, and flexibility of the robots. Inspired by the hydraulic joints of spider leg, an innovative folding membrane soft joint actuator is designed to achieve an 80° rotation angle at 5.25 kPa air pressure, with an angle density of 15.23°/ kPa and a fatigue life of more than 10,000 times, significantly improving the output characteristics of pneumatic joints. This actuator's folding membrane is made of silicone-coated fiber fabric and is manufactured via integrated molding method. Furthermore, regarding the issue of ineffective expansion in actuators, the effects of various constraint layer materials on the actuator's output performance are investigated and discussed. It is found that the TPU-constrained actuator has a maximum output torque of 0.307 N·m, which is 7.7% and 13.7% greater than that of the fiber-constrained and unconstrained actuators, respectively. Meanwhile, the rebound time was 1.25 seconds, which was reduced by 0.23 seconds and 0.22 seconds respectively. Soft actuators can benefit from the design technique of the folding membrane soft joint actuator presented in the study, which can provide significant torque output at low air pressure and has superior fatigue characteristics.","PeriodicalId":13241,"journal":{"name":"IEEE Robotics and Automation Letters","volume":"10 2","pages":"1401-1408"},"PeriodicalIF":4.6,"publicationDate":"2024-12-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142912526","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 : 2024-12-23DOI: 10.1109/LRA.2024.3520919
Aniket Datar;Chenhui Pan;Xuesu Xiao
Most autonomous navigation systems assume wheeled robots are rigid bodies and their 2D planar workspaces can be divided into free spaces and obstacles. However, recent wheeled mobility research, showing that wheeled platforms have the potential of moving over vertically challenging terrain (e.g., rocky outcroppings, rugged boulders, and fallen tree trunks), invalidate both assumptions. Navigating off-road vehicle chassis with long suspension travel and low tire pressure in places where the boundary between obstacles and free spaces is blurry requires precise 3D modeling of the interaction between the chassis and the terrain, which is complicated by suspension and tire deformation, varying tire-terrain friction, vehicle weight distribution and momentum, etc. In this letter, we present a learning approach to model wheeled mobility, i.e., in terms of vehicle-terrain forward dynamics, and plan feasible, stable, and efficient motion to drive over vertically challenging terrain without rolling over or getting stuck. We present physical experiments on two wheeled robots and show that planning using our learned model can achieve up to 60% improvement in navigation success rate and 46% reduction in unstable chassis roll and pitch angles.
{"title":"Learning to Model and Plan for Wheeled Mobility on Vertically Challenging Terrain","authors":"Aniket Datar;Chenhui Pan;Xuesu Xiao","doi":"10.1109/LRA.2024.3520919","DOIUrl":"https://doi.org/10.1109/LRA.2024.3520919","url":null,"abstract":"Most autonomous navigation systems assume wheeled robots are rigid bodies and their 2D planar workspaces can be divided into free spaces and obstacles. However, recent wheeled mobility research, showing that wheeled platforms have the potential of moving over vertically challenging terrain (e.g., rocky outcroppings, rugged boulders, and fallen tree trunks), invalidate both assumptions. Navigating off-road vehicle chassis with long suspension travel and low tire pressure in places where the boundary between obstacles and free spaces is blurry requires precise 3D modeling of the interaction between the chassis and the terrain, which is complicated by suspension and tire deformation, varying tire-terrain friction, vehicle weight distribution and momentum, etc. In this letter, we present a learning approach to model wheeled mobility, i.e., in terms of vehicle-terrain forward dynamics, and plan feasible, stable, and efficient motion to drive over vertically challenging terrain without rolling over or getting stuck. We present physical experiments on two wheeled robots and show that planning using our learned model can achieve up to 60% improvement in navigation success rate and 46% reduction in unstable chassis roll and pitch angles.","PeriodicalId":13241,"journal":{"name":"IEEE Robotics and Automation Letters","volume":"10 2","pages":"1505-1512"},"PeriodicalIF":4.6,"publicationDate":"2024-12-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142938312","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 : 2024-12-23DOI: 10.1109/LRA.2024.3521181
Tao Li;Zhenbao Yu;Banglei Guan;Jianli Han;Weimin Lv;Friedrich Fraundorfer
This work presents two novel solvers for estimating the relative poses among views with known vertical directions. The vertical directions of camera views can be easily obtained using inertial measurement units (IMUs) which have been widely used in autonomous vehicles, mobile phones, and autonomous aerial vehicles (AAVs). Given the known vertical directions, our algorithms only need to solve for two rotation angles and two translation vectors. In this paper, a linear closed-form solution has been described, requiring only four point correspondences in three views. We also propose a minimal solution with three point correspondences using the latest Gröbner basis solver. Since the proposed methods require fewer point correspondences, they can be efficiently applied within the RANSAC framework for outliers removal and pose estimation in visual odometry. The proposed method has been tested on both synthetic data and real-world scenes from KITTI. The experimental results show that the accuracy of the estimated poses is superior to other alternative methods.
{"title":"Trifocal Tensor and Relative Pose Estimation With Known Vertical Direction","authors":"Tao Li;Zhenbao Yu;Banglei Guan;Jianli Han;Weimin Lv;Friedrich Fraundorfer","doi":"10.1109/LRA.2024.3521181","DOIUrl":"https://doi.org/10.1109/LRA.2024.3521181","url":null,"abstract":"This work presents two novel solvers for estimating the relative poses among views with known vertical directions. The vertical directions of camera views can be easily obtained using inertial measurement units (IMUs) which have been widely used in autonomous vehicles, mobile phones, and autonomous aerial vehicles (AAVs). Given the known vertical directions, our algorithms only need to solve for two rotation angles and two translation vectors. In this paper, a linear closed-form solution has been described, requiring only four point correspondences in three views. We also propose a minimal solution with three point correspondences using the latest Gröbner basis solver. Since the proposed methods require fewer point correspondences, they can be efficiently applied within the RANSAC framework for outliers removal and pose estimation in visual odometry. The proposed method has been tested on both synthetic data and real-world scenes from KITTI. The experimental results show that the accuracy of the estimated poses is superior to other alternative methods.","PeriodicalId":13241,"journal":{"name":"IEEE Robotics and Automation Letters","volume":"10 2","pages":"1305-1312"},"PeriodicalIF":4.6,"publicationDate":"2024-12-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142912515","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 : 2024-12-23DOI: 10.1109/LRA.2024.3520920
Paolo Forte;Himanshu Gupta;Henrik Andreasson;Uwe Köckemann;Achim J. Lilienthal
We propose a context-aware navigation framework designed to support the navigation of autonomous ground vehicles, including articulated ones. The proposed framework employs a behavior tree with novel nodes to manage the navigation tasks: planner and controller selections, path planning, path following, and recovery. It incorporates a weather detection system and configurable global path planning and controller strategy selectors implemented as behavior tree action nodes. These components are integrated into a sub-tree that supervises and manages available options and parameters for global planners and control strategies by evaluating map and real-time sensor data. The proposed approach offers three key benefits: overcoming the limitations of single planner strategies in challenging scenarios; ensuring efficient path planning by balancing between optimization and computational effort; and achieving smoother navigation by reducing path curvature and improving drivability. The performance of the proposed framework is analyzed empirically, and compared against state of the art navigation systems with single path planning strategies.
{"title":"On Robust Context-Aware Navigation for Autonomous Ground Vehicles","authors":"Paolo Forte;Himanshu Gupta;Henrik Andreasson;Uwe Köckemann;Achim J. Lilienthal","doi":"10.1109/LRA.2024.3520920","DOIUrl":"https://doi.org/10.1109/LRA.2024.3520920","url":null,"abstract":"We propose a context-aware navigation framework designed to support the navigation of autonomous ground vehicles, including articulated ones. The proposed framework employs a behavior tree with novel nodes to manage the navigation tasks: planner and controller selections, path planning, path following, and recovery. It incorporates a weather detection system and configurable global path planning and controller strategy selectors implemented as behavior tree action nodes. These components are integrated into a sub-tree that supervises and manages available options and parameters for global planners and control strategies by evaluating map and real-time sensor data. The proposed approach offers three key benefits: overcoming the limitations of single planner strategies in challenging scenarios; ensuring efficient path planning by balancing between optimization and computational effort; and achieving smoother navigation by reducing path curvature and improving drivability. The performance of the proposed framework is analyzed empirically, and compared against state of the art navigation systems with single path planning strategies.","PeriodicalId":13241,"journal":{"name":"IEEE Robotics and Automation Letters","volume":"10 2","pages":"1449-1456"},"PeriodicalIF":4.6,"publicationDate":"2024-12-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10812052","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142918211","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}
Reinforcement learning (RL) has shown great effectiveness in quadrotor control, enabling specialized policies to develop even human-champion-level performance in single-task scenarios. However, these specialized policies often struggle with novel tasks, requiring a complete retraining of the policy from scratch. To address this limitation, this paper presents a novel multi-task reinforcement learning (MTRL) framework tailored for quadrotor control, leveraging the shared physical dynamics of the platform to enhance sample efficiency and task performance. By employing a multi-critic architecture and shared task encoders, our framework facilitates knowledge transfer across tasks, enabling a single policy to execute diverse maneuvers, including high-speed stabilization, velocity tracking, and autonomous racing. Our experimental results, validated both in simulation and real-world scenarios, demonstrate that our framework outperforms baseline approaches in terms of sample efficiency and overall task performance. Video is available at https://youtu.be/HfK9UT1OVnY.
{"title":"Multi-Task Reinforcement Learning for Quadrotors","authors":"Jiaxu Xing;Ismail Geles;Yunlong Song;Elie Aljalbout;Davide Scaramuzza","doi":"10.1109/LRA.2024.3520894","DOIUrl":"https://doi.org/10.1109/LRA.2024.3520894","url":null,"abstract":"Reinforcement learning (RL) has shown great effectiveness in quadrotor control, enabling specialized policies to develop even human-champion-level performance in single-task scenarios. However, these specialized policies often struggle with novel tasks, requiring a complete retraining of the policy from scratch. To address this limitation, this paper presents a novel multi-task reinforcement learning (MTRL) framework tailored for quadrotor control, leveraging the shared physical dynamics of the platform to enhance sample efficiency and task performance. By employing a multi-critic architecture and shared task encoders, our framework facilitates knowledge transfer across tasks, enabling a single policy to execute diverse maneuvers, including high-speed stabilization, velocity tracking, and autonomous racing. Our experimental results, validated both in simulation and real-world scenarios, demonstrate that our framework outperforms baseline approaches in terms of sample efficiency and overall task performance. Video is available at <uri>https://youtu.be/HfK9UT1OVnY</uri>.","PeriodicalId":13241,"journal":{"name":"IEEE Robotics and Automation Letters","volume":"10 3","pages":"2112-2119"},"PeriodicalIF":4.6,"publicationDate":"2024-12-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142993213","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 : 2024-12-23DOI: 10.1109/LRA.2024.3520916
Daniel Sliwowski;Dongheui Lee
The introduction of robots into everyday scenarios necessitates algorithms capable of monitoring the execution of tasks. In this letter, we propose ConditionNET, an approach for learning the preconditions and effects of actions in a fully data-driven manner. We develop an efficient vision-language model and introduce additional optimization objectives during training to optimize for consistent feature representations. ConditionNET explicitly models the dependencies between actions, preconditions, and effects, leading to improved performance. We evaluate our model on two robotic datasets, one of which we collected for this letter, containing 406 successful and 138 failed teleoperated demonstrations of a Franka Emika Panda robot performing tasks like pouring and cleaning the counter. We show in our experiments that ConditionNET outperforms all baselines on both anomaly detection and phase prediction tasks. Furthermore, we implement an action monitoring system on a real robot to demonstrate the practical applicability of the learned preconditions and effects. Our results highlight the potential of ConditionNET for enhancing the reliability and adaptability of robots in real-world environments.
{"title":"ConditionNET: Learning Preconditions and Effects for Execution Monitoring","authors":"Daniel Sliwowski;Dongheui Lee","doi":"10.1109/LRA.2024.3520916","DOIUrl":"https://doi.org/10.1109/LRA.2024.3520916","url":null,"abstract":"The introduction of robots into everyday scenarios necessitates algorithms capable of monitoring the execution of tasks. In this letter, we propose ConditionNET, an approach for learning the preconditions and effects of actions in a fully data-driven manner. We develop an efficient vision-language model and introduce additional optimization objectives during training to optimize for consistent feature representations. ConditionNET explicitly models the dependencies between actions, preconditions, and effects, leading to improved performance. We evaluate our model on two robotic datasets, one of which we collected for this letter, containing 406 successful and 138 failed teleoperated demonstrations of a Franka Emika Panda robot performing tasks like pouring and cleaning the counter. We show in our experiments that ConditionNET outperforms all baselines on both anomaly detection and phase prediction tasks. Furthermore, we implement an action monitoring system on a real robot to demonstrate the practical applicability of the learned preconditions and effects. Our results highlight the potential of ConditionNET for enhancing the reliability and adaptability of robots in real-world environments.","PeriodicalId":13241,"journal":{"name":"IEEE Robotics and Automation Letters","volume":"10 2","pages":"1337-1344"},"PeriodicalIF":4.6,"publicationDate":"2024-12-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142912458","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}
Roadside camera-based perception methods are in high demand for developing efficient vehicle-infrastructure collaborative perception systems. By focusing on object-level depth prediction, we explore the potential benefits of integrating environmental priors into such systems and propose a geometry-based roadside per-object depth estimation algorithm dubbed GARD. The proposed method capitalizes on the inherent geometric properties of the pinhole camera model to derive depth as well as 3D positions for given 2D targets in roadside-view images, alleviating the need for computationally intensive end-to-end learning architectures for monocular 3D detection. Using only a pre-trained 2D detection model, our approach does not require vast amounts of scene-specific training data and shows superior generalization abilities across varying environments and camera setups, making it a practical and cost-effective solution for monocular 3D object detection.
{"title":"GARD: A Geometry-Informed and Uncertainty-Aware Baseline Method for Zero-Shot Roadside Monocular Object Detection","authors":"Yuru Peng;Beibei Wang;Zijian Yu;Lu Zhang;Jianmin Ji;Yu Zhang;Yanyong Zhang","doi":"10.1109/LRA.2024.3520923","DOIUrl":"https://doi.org/10.1109/LRA.2024.3520923","url":null,"abstract":"Roadside camera-based perception methods are in high demand for developing efficient vehicle-infrastructure collaborative perception systems. By focusing on object-level depth prediction, we explore the potential benefits of integrating environmental priors into such systems and propose a geometry-based roadside per-object depth estimation algorithm dubbed GARD. The proposed method capitalizes on the inherent geometric properties of the pinhole camera model to derive depth as well as 3D positions for given 2D targets in roadside-view images, alleviating the need for computationally intensive end-to-end learning architectures for monocular 3D detection. Using only a pre-trained 2D detection model, our approach does not require vast amounts of scene-specific training data and shows superior generalization abilities across varying environments and camera setups, making it a practical and cost-effective solution for monocular 3D object detection.","PeriodicalId":13241,"journal":{"name":"IEEE Robotics and Automation Letters","volume":"10 2","pages":"1297-1304"},"PeriodicalIF":4.6,"publicationDate":"2024-12-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142912470","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 : 2024-12-20DOI: 10.1109/LRA.2024.3520436
Alan Li;Angela P. Schoellig
6D Object pose estimation is a fundamental component in robotics enabling efficient interaction with the environment. It is particularly challenging in bin-picking applications, where objects may be textureless and in difficult poses, and occlusion between objects of the same type may cause confusion even in well-trained models. We propose a novel method of hard example synthesis that is model-agnostic, using existing simulators and the modeling of pose error in both the camera-to-object viewsphere and occlusion space. Through evaluation of the model performance with respect to the distribution of object poses and occlusions, we discover regions of high error and generate realistic training samples to specifically target these regions. With our training approach, we demonstrate an improvement in correct detection rate of up to 20% across several ROBI-dataset objects using state-of-the-art pose estimation models.
{"title":"Targeted Hard Sample Synthesis Based on Estimated Pose and Occlusion Error for Improved Object Pose Estimation","authors":"Alan Li;Angela P. Schoellig","doi":"10.1109/LRA.2024.3520436","DOIUrl":"https://doi.org/10.1109/LRA.2024.3520436","url":null,"abstract":"6D Object pose estimation is a fundamental component in robotics enabling efficient interaction with the environment. It is particularly challenging in bin-picking applications, where objects may be textureless and in difficult poses, and occlusion between objects of the same type may cause confusion even in well-trained models. We propose a novel method of hard example synthesis that is model-agnostic, using existing simulators and the modeling of pose error in both the camera-to-object viewsphere and occlusion space. Through evaluation of the model performance with respect to the distribution of object poses and occlusions, we discover regions of high error and generate realistic training samples to specifically target these regions. With our training approach, we demonstrate an improvement in correct detection rate of up to 20% across several ROBI-dataset objects using state-of-the-art pose estimation models.","PeriodicalId":13241,"journal":{"name":"IEEE Robotics and Automation Letters","volume":"10 2","pages":"1281-1288"},"PeriodicalIF":4.6,"publicationDate":"2024-12-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142912366","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 : 2024-12-20DOI: 10.1109/LRA.2024.3521178
Nick Ah Sen;Dana Kulić;Pamela Carreno-Medrano
In shared human-robot environments, effective navigation requires robots to adapt to various pedestrian behaviors encountered in the real world. Most existing deep reinforcement learning algorithms for human-aware robot navigation typically assume that pedestrians adhere to a single walking behavior during training, limiting their practicality/performance in scenarios where pedestrians exhibit various types of behavior. In this work, we propose to enhance the generalization capabilities of human-aware robot navigation by employing Domain Randomization (DR) techniques to train navigation policies on a diverse range of simulated pedestrian behaviors with the hope of better generalization to the real world. We evaluate the effectiveness of our method by comparing the generalization capabilities of a robot navigation policy trained with and without DR, both in simulations and through a real-user study, focusing on adaptability to different pedestrian behaviors, performance in novel environments, and users' perceived comfort, sociability and naturalness. Our findings reveal that the use of DR significantly enhances the robot's social compliance in both simulated and real-life contexts.
{"title":"Domain Randomization for Learning to Navigate in Human Environments","authors":"Nick Ah Sen;Dana Kulić;Pamela Carreno-Medrano","doi":"10.1109/LRA.2024.3521178","DOIUrl":"https://doi.org/10.1109/LRA.2024.3521178","url":null,"abstract":"In shared human-robot environments, effective navigation requires robots to adapt to various pedestrian behaviors encountered in the real world. Most existing deep reinforcement learning algorithms for human-aware robot navigation typically assume that pedestrians adhere to a single walking behavior during training, limiting their practicality/performance in scenarios where pedestrians exhibit various types of behavior. In this work, we propose to enhance the generalization capabilities of human-aware robot navigation by employing Domain Randomization (DR) techniques to train navigation policies on a diverse range of simulated pedestrian behaviors with the hope of better generalization to the real world. We evaluate the effectiveness of our method by comparing the generalization capabilities of a robot navigation policy trained with and without DR, both in simulations and through a real-user study, focusing on adaptability to different pedestrian behaviors, performance in novel environments, and users' perceived comfort, sociability and naturalness. Our findings reveal that the use of DR significantly enhances the robot's social compliance in both simulated and real-life contexts.","PeriodicalId":13241,"journal":{"name":"IEEE Robotics and Automation Letters","volume":"10 2","pages":"1625-1632"},"PeriodicalIF":4.6,"publicationDate":"2024-12-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142937838","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 : 2024-12-20DOI: 10.1109/LRA.2024.3521179
Huazhang Zhu;Tian Lan;Shunzheng Ma;Xuan Zhao;Huiliang Shang;Ruijiao Li
We present an autonomous exploration method for autonomous aerial vehicles (AAVs) for three-dimensional (3D) exploration tasks. Our approach, utilizing a cooperation strategy between common viewpoints and frontier viewpoints, fully leverages the agility and flexibility of AAVs, demonstrating faster and more comprehensive exploration than the current state-of-the-art. Common viewpoints, specifically designed for AAVs exploration, are evenly distributed throughout the 3D space for 3D exploration tasks. Frontier viewpoints are positioned at the centroids of clusters of frontier points to help the AAV maintain motivation to explore unknown complex 3D environments and navigate through narrow corners and passages. This strategy allows the AAV to access every corner of the 3D environment. Additionally, our method includes a refined relocation mechanism for AAVs specifically. Experimental comparisons show that our method ensures complete exploration coverage in environments with complex terrain. Our method outperforms TARE DSVP, GBP and MBP by the coverage rate of 64%, 63%, 54% and 49% respectively in garage-D. In narrow tunnels, ours and DSVP are the only two evaluated methods that achieve complete coverage, with ours outperforming DSVP by 35% in exploration efficiency.
{"title":"CODE: Complete Coverage AAV Exploration Planner Using Dual-Type Viewpoints for Multi-Layer Complex Environments","authors":"Huazhang Zhu;Tian Lan;Shunzheng Ma;Xuan Zhao;Huiliang Shang;Ruijiao Li","doi":"10.1109/LRA.2024.3521179","DOIUrl":"https://doi.org/10.1109/LRA.2024.3521179","url":null,"abstract":"We present an autonomous exploration method for autonomous aerial vehicles (AAVs) for three-dimensional (3D) exploration tasks. Our approach, utilizing a cooperation strategy between common viewpoints and frontier viewpoints, fully leverages the agility and flexibility of AAVs, demonstrating faster and more comprehensive exploration than the current state-of-the-art. Common viewpoints, specifically designed for AAVs exploration, are evenly distributed throughout the 3D space for 3D exploration tasks. Frontier viewpoints are positioned at the centroids of clusters of frontier points to help the AAV maintain motivation to explore unknown complex 3D environments and navigate through narrow corners and passages. This strategy allows the AAV to access every corner of the 3D environment. Additionally, our method includes a refined relocation mechanism for AAVs specifically. Experimental comparisons show that our method ensures complete exploration coverage in environments with complex terrain. Our method outperforms TARE DSVP, GBP and MBP by the coverage rate of 64%, 63%, 54% and 49% respectively in garage-D. In narrow tunnels, ours and DSVP are the only two evaluated methods that achieve complete coverage, with ours outperforming DSVP by 35% in exploration efficiency.","PeriodicalId":13241,"journal":{"name":"IEEE Robotics and Automation Letters","volume":"10 2","pages":"1880-1887"},"PeriodicalIF":4.6,"publicationDate":"2024-12-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142975964","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：