Frontiers in Robotics and AI最新文献

{"title":"Dense mapping from sparse visual odometry: a lightweight uncertainty-guaranteed depth completion method.","authors":"Daolong Yang, Xudong Zhang, Haoyuan Liu, Haoyang Wu, Chengcai Wang, Kun Xu, Xilun Ding","doi":"10.3389/frobt.2025.1644230","DOIUrl":"10.3389/frobt.2025.1644230","url":null,"abstract":"<p><strong>Introduction: </strong>Visual odometry (VO) has been widely deployed on mobile robots for spatial perception. State-of-the-art VO offers robust localization, the maps it generates are often too sparse for downstream tasks due to insufffcient depth data. While depth completion methods can estimate dense depth from sparse data, the extreme sparsity and highly uneven distribution of depth signals in VO (∼ 0.15% of the pixels in the depth image available) poses signiffcant challenges.</p><p><strong>Methods: </strong>To address this issue, we propose a lightweight Image-Guided Uncertainty-Aware Depth Completion Network (IU-DC) for completing sparse depth from VO. This network integrates color and spatial information into a normalized convolutional neural network to tackle the sparsity issue and simultaneously outputs dense depth and associated uncertainty. The estimated depth is uncertainty-aware, allowing for the filtering of outliers and ensuring precise spatial perception.</p><p><strong>Results: </strong>The superior performance of IU-DC compared to SOTA is validated across multiple open-source datasets in terms of depth and uncertainty estimation accuracy. In real-world mapping tasks, by integrating IU-DC with the mapping module, we achieve 50 × more reconstructed volumes and 78% coverage of the ground truth with twice the accuracy compared to SOTA, despite having only 0.6 M parameters (just 3% of the size of the SOTA).</p><p><strong>Discussion: </strong>Our code will be released at https://github.com/YangDL-BEIHANG/Dense-mapping-from-sparse-visual-odometry/tree/d5a11b4403b5ac2e9e0c3644b14b9711c2748bf9.</p>","PeriodicalId":47597,"journal":{"name":"Frontiers in Robotics and AI","volume":"12 ","pages":"1644230"},"PeriodicalIF":3.0,"publicationDate":"2025-09-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12497602/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145245501","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"GNV2-SLAM: vision SLAM system for cowshed inspection robots.","authors":"Xinwu Du, Tingting Li, Xin Jin, Xiufang Yu, Xiaolin Xie, Chenglin Zhang","doi":"10.3389/frobt.2025.1648309","DOIUrl":"10.3389/frobt.2025.1648309","url":null,"abstract":"<p><p>Simultaneous Localization and Mapping (SLAM) has emerged as one of the foundational technologies enabling mobile robots to achieve autonomous navigation, garnering significant attention in recent years. To address the limitations inherent in traditional SLAM systems when operating within dynamic environments, this paper proposes a new SLAM system named GNV2-SLAM based on ORB-SLAM2, offering an innovative solution for the scenario of cowshed inspection. This innovative system incorporates a lightweight object detection network called GNV2 based on YOLOv8. Additionally, it employs GhostNetv2 as backbone network. The CBAM attention mechanism and SCDown downsampling module were introduced to reduce the model complexity while ensuring detection accuracy. Experimental results indicate that the GNV2 network achieves excellent model compression effects while maintaining high performance: mAP@0.5 increased by 1.04%, reaching a total of 95.19%; model parameters were decreased by 41.95%, computational cost reduced by 36.71%, and the model size shrunk by 40.44%. Moreover, the GNV2-SLAM system incorporates point and line feature extraction techniques, effectively mitigate issues reduced feature point extraction caused by excessive dynamic targets or blurred images. Testing on the TUM dataset demonstrate that GNV2-SLAM significantly outperforms the traditional ORB-SLAM2 system in terms of positioning accuracy and robustness within dynamic environments. Specifically, there was a remarkable reduction of 96.13% in root mean square error (RMSE) for absolute trajectory error (ATE), alongside decreases of 88.36% and 86.19% for translation and rotation drift in relative pose error (RPE), respectively. In terms of tracking evaluation, GNV2-SLAM successfully completes the tracking processing of a single frame image within 30 ms, demonstrating expressive real-time performance and competitiveness. Following the deployment of this system on inspection robots and subsequent experimental trials conducted in the cowshed environment, the results indicate that when the robot operates at speeds of 0.4 m/s and 0.6 m/s, the pose trajectory output by GNV2-SLAM is more consistent with the robot's actual movement trajectory. This study systematically validated the system's significant advantages in target recognition and positioning accuracy through experimental verification, thereby providing a new technical solution for the comprehensive automation of cattle barn inspection tasks.</p>","PeriodicalId":47597,"journal":{"name":"Frontiers in Robotics and AI","volume":"12 ","pages":"1648309"},"PeriodicalIF":3.0,"publicationDate":"2025-09-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12492984/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145233786","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Autonomy in socially assistive robotics: a systematic review.","authors":"Romain Maure, Barbara Bruno","doi":"10.3389/frobt.2025.1586473","DOIUrl":"10.3389/frobt.2025.1586473","url":null,"abstract":"<p><p>Socially assistive robots are increasingly being researched and deployed in various domains such as education, healthcare, service, and even as collaborators in a variety of other workplaces. Similarly, SARs are also expected to interact in a socially acceptable manner with a wide audience, ranging from preschool children to the elderly. This diversity of application domains and target populations raises technical and social challenges that are yet to be overcome. While earlier works relied on the Wizard-of-Oz (WoZ) paradigm to give an illusion of interactivity and intelligence, a transition toward more autonomous robots can be observed. In this article, we present a systematic review, following the PRISMA method, of the last 5 years of Socially Assistive Robotics research, centered around SARs' level of autonomy with a stronger focus on fully and semi-autonomous robots than non-autonomous ones. Specifically, to analyse SARs' level of autonomy, the review identifies which sensing and actuation capabilities of SARs are typically automated and which ones are not, and how these capabilities are automated, with the aim of identifying potential gaps to be explored in future research. The review further explores whether SARs' level of autonomy and capabilities are transparently communicated to the diverse target audiences above described and discusses the potential benefits and drawbacks of such transparency. Finally, with the aim of providing a more holistic view of SARs' characteristics and application domains, the review also reports the embodiment and commonly envisioned role of SARs, as well as their interventions' size, length and environment.</p>","PeriodicalId":47597,"journal":{"name":"Frontiers in Robotics and AI","volume":"12 ","pages":"1586473"},"PeriodicalIF":3.0,"publicationDate":"2025-09-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12491022/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145233743","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Correction: Understanding consumer attitudes towards second-hand robots for the home.","authors":"Helen McGloin, Matthew Studley, Richard Mawle, Alan Frank Thomas Winfield","doi":"10.3389/frobt.2025.1694690","DOIUrl":"https://doi.org/10.3389/frobt.2025.1694690","url":null,"abstract":"<p><p>[This corrects the article DOI: 10.3389/frobt.2024.1324519.].</p>","PeriodicalId":47597,"journal":{"name":"Frontiers in Robotics and AI","volume":"12 ","pages":"1694690"},"PeriodicalIF":3.0,"publicationDate":"2025-09-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12489513/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145233723","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"VIO-GO: optimizing event-based SLAM parameters for robust performance in high dynamic range scenarios.","authors":"Saber Sakhrieh, Abhilasha Singh, Jinane Mounsef, Bilal Arain, Noel Maalouf","doi":"10.3389/frobt.2025.1541017","DOIUrl":"10.3389/frobt.2025.1541017","url":null,"abstract":"<p><p>This paper addresses a critical challenge in Industry 4.0 robotics by enhancing Visual Inertial Odometry (VIO) systems to operate effectively in dynamic and low-light industrial environments, which are common in sectors like warehousing, logistics, and manufacturing. Inspired by biological sensing mechanisms, we integrate bio-inspired event cameras to improve state estimation systems performance in both dynamic and low-light conditions, enabling reliable localization and mapping. The proposed state estimation framework integrates events, conventional video frames, and inertial data to achieve reliable and precise localization with specific emphasis on real-world challenges posed by high-speed and cluttered settings typical in Industry 4.0. Despite advancements in event-based sensing, there is a noteworthy gap in optimizing Event Simultaneous Localization and Mapping (SLAM) parameters for practical applications. To address this, we introduce a novel VIO-Gradient-based Optimization (VIO-GO) method that employs Batch Gradient Descent (BGD) for efficient parameter tuning. This automated approach determines optimal parameters for Event SLAM algorithms by using motion-compensated images to represent event data. Experimental validation on the Event Camera Dataset shows a remarkable 60% improvement in Mean Position Error (MPE) over fixed-parameter methods. Our results demonstrate that VIO-GO consistently identifies optimal parameters, enabling precise VIO performance in complex, dynamic scenarios essential for Industry 4.0 applications. Additionally, as parameter complexity scales, VIO-GO achieves a 24% reduction in MPE when using the most comprehensive parameter set (VIO-GO8) compared to a minimal set (VIO-GO2), highlighting the method's scalability and robustness for adaptive robotic systems in challenging industrial environments.</p>","PeriodicalId":47597,"journal":{"name":"Frontiers in Robotics and AI","volume":"12 ","pages":"1541017"},"PeriodicalIF":3.0,"publicationDate":"2025-09-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12490131/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145233765","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A PSO-ML-LSTM-based IMU state estimation approach for manipulator teleoperation.","authors":"Renyi Zhou, Yuanchong Li, Aimin Zhang, Tie Zhang, Yisheng Guan, Zhijia Zhao, Shouyan Chen","doi":"10.3389/frobt.2025.1638853","DOIUrl":"10.3389/frobt.2025.1638853","url":null,"abstract":"<p><p>Manipulator teleoperation can liberate humans from hazardous tasks. Signal noise caused by environmental disturbances and the devices' inherent characteristics may limit the teleoperation performance. This paper proposes an approach for inertial measurement unit (IMU) state estimation based on particle swarm optimization (PSO) and modulated long short-term memory (ML-LSTM) neural networks to mitigate the impact of IMU cumulative error on the robot teleoperation performance. A motion mapping model for the human arm and a seven-degree-of-freedom (7-DOF) robotic arm are first established based on global configuration parameters and a hybrid mapping method. This model is used to describe the impact of IMU cumulative error on the robot teleoperation performance. Subsequently, the IMU pose state estimation model is constructed using PSO and ML-LSTM neural networks. The initial data of multiple IMUs and handling handles are used for training the estimation model. Finally, comparative experiments are conducted to verify the performance of the proposed state estimation model. The results demonstrate that the PSO-ML-LSTM algorithm can effectively eliminate the impact of IMU cumulative errors on robot teleoperation.</p>","PeriodicalId":47597,"journal":{"name":"Frontiers in Robotics and AI","volume":"12 ","pages":"1638853"},"PeriodicalIF":3.0,"publicationDate":"2025-09-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12481027/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145208004","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Gödelian embodied self-referential genomic intelligence: lessons for AI and AGI from the genomic blockchain.","authors":"Sheri Markose","doi":"10.3389/frobt.2025.1624695","DOIUrl":"10.3389/frobt.2025.1624695","url":null,"abstract":"<p><p>The security of code-based digital records is a major concern of the 21st century. AI and artificial general intelligence (AGI) can be hacked to pieces by digital adversaries, and some AI objectives can lead to existential threats. The former arises from sitting duck problems that all software systems are vulnerable to, and the latter include control and misalignment problems. Blockchain technology, circa 2009, can address these problems: hashing algorithms rely on a consensus mechanism in manmade software systems to keep early blocks of software immutable and tamper-proof from digital malware, while new blocks can be added only if consistently aligned with original blocks. There is evidence that the ancient precedent of the genomic blockchain, underpinning the unbroken chain of life, uses a self-referential rather than a consensus-based hashing algorithm. Knowledge of self-codes permits biotic elements to achieve a hack-free agenda by self-reporting that they have been \"negated,\" or hacked, exactly implementing the Gödel sentence from foundational mathematics of Gödel, Turing, and Post (G-T-P). This results in an arms race in open-ended novelty to secure the primacy of original self-codes. Selfhood and autonomy are staples of neuroscience on complex self-other social cognition and increasingly of autonomous AGI agents capable of end-to-end programmed self-assembly. My perspective is that self-referential G-T-P information processing, first found in the adaptive immune system of jawed fish 500 mya and more recently in mirror neuron systems of humans, has enabled code-based self-organized intelligent systems like life to survive over 3.7 billion years. Some lessons for AGI can be gleaned from this discussion.</p>","PeriodicalId":47597,"journal":{"name":"Frontiers in Robotics and AI","volume":"12 ","pages":"1624695"},"PeriodicalIF":3.0,"publicationDate":"2025-09-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12477408/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145201637","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A comprehensive review and bibliometric analysis on collaborative robotics for industry: safety emerging as a core focus.","authors":"Aida Haghighi, Morteza Cheraghi, Jérôme Pocachard, Valérie Botta-Genoulaz, Sabrina Jocelyn, Hamidreza Pourzarei","doi":"10.3389/frobt.2025.1605682","DOIUrl":"10.3389/frobt.2025.1605682","url":null,"abstract":"<p><p>Research organizations and academics often seek to map the development of scientific fields, identify research gaps, and guide the direction of future research. In cobot-related research, the scientific literature consulted does not propose any comprehensive research agenda. Moreover, cobots, industrial robots inherently designed to collaborate with humans, bring with them emerging issues. To solve them, interdisciplinary research is often essential (e.g., combination of engineering, ergonomics and biomechanics expertise to handle safety challenges). This paper proposes an exhaustive study that employs a scoping review and bibliometric analysis to provide a structured macro perspective on the developments, key topics, and trends in cobot research for industry. A total of 2,195 scientific publications were gained from the Web of Science database, and a thorough selection process narrowed them down to 532 papers for comprehensive analysis. Descriptive statistics were employed to analyze bibliometric measures, highlighting publication trends, leading journals, the most productive institutions, engaged countries, influential authors, and prominent research topics. Co-authorship and bibliographic couplings were also examined. Through a co-occurrence analysis of terms, the content and research objectives of the papers were systematically reviewed and lead to a univocal categorization framework. That categorization can support organizations or researchers in different cobotics (collaborative robotics) fields by understanding research developments and trends, identifying collaboration opportunities, selecting suitable publication venues, advancing the theoretical and experimental understanding of automatic collaborative systems, and identifying research directions and predicting the evolution of publication quantity in cobotics.</p>","PeriodicalId":47597,"journal":{"name":"Frontiers in Robotics and AI","volume":"12 ","pages":"1605682"},"PeriodicalIF":3.0,"publicationDate":"2025-09-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12464494/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145187143","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A projection-based inverse kinematic model for extensible continuum robots and hyper-redundant robots with an elbow joint.","authors":"Sven Fritsch, Dirk Oberschmidt","doi":"10.3389/frobt.2025.1627688","DOIUrl":"10.3389/frobt.2025.1627688","url":null,"abstract":"<p><p>Inverse kinematics is a core problem in robotics, involving the use of kinematic equations to calculate the joint configurations required to achieve a target pose. This study introduces a novel inverse kinematic model (IKM) for extensible (i.e., length-adjustable) continuum robots (CRs) and hyper-redundant robots (HRRs) featuring an elbow joint. This IKM numerically solves a set of equations representing geometric constraints (abbreviated as NSGC). NSGC can handle target poses <math> <mrow> <msub><mrow><mi>X</mi></mrow> <mrow><mi>t</mi></mrow> </msub> <mo>=</mo> <mrow><mo>[</mo> <mrow> <msub><mrow><mi>x</mi></mrow> <mrow><mi>t</mi></mrow> </msub> <mo>,</mo> <msub><mrow><mi>y</mi></mrow> <mrow><mi>t</mi></mrow> </msub> <mo>,</mo> <msub><mrow><mi>z</mi></mrow> <mrow><mi>t</mi></mrow> </msub> <mo>,</mo> <msub><mrow><mi>ψ</mi></mrow> <mrow><mi>t</mi></mrow> </msub> </mrow> <mo>]</mo></mrow> </mrow> </math> in 3D space, which are projected onto a 2D plane and solved numerically. NSGC is capable of real-time operation and accounts for elbow joint limits. Extensive simulations and empirical tests confirm the reliability, performance, and practical applicability of NSGC.</p>","PeriodicalId":47597,"journal":{"name":"Frontiers in Robotics and AI","volume":"12 ","pages":"1627688"},"PeriodicalIF":3.0,"publicationDate":"2025-09-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12464886/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145187126","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Enhancing weed detection through knowledge distillation and attention mechanism.","authors":"Ali El Alaoui, Hajar Mousannif","doi":"10.3389/frobt.2025.1654074","DOIUrl":"10.3389/frobt.2025.1654074","url":null,"abstract":"<p><p>Weeds pose a significant challenge in agriculture by competing with crops for essential resources, leading to reduced yields. To address this issue, researchers have increasingly adopted advanced machine learning techniques. Recently, Vision Transformers (ViT) have demonstrated remarkable success in various computer vision tasks, making their application to weed classification, detection, and segmentation more advantageous compared to traditional Convolutional Neural Networks (CNNs) due to their self-attention mechanism. However, the deployment of these models in agricultural robotics is hindered by resource limitations. Key challenges include high training costs, the absence of inductive biases, the extensive volume of data required for training, model size, and runtime memory constraints. This study proposes a knowledge distillation-based method for optimizing the ViT model. The approach aims to enhance the ViT model architecture while maintaining its performance for weed detection. To facilitate the training of the compacted ViT student model and enable parameter sharing and local receptive fields, knowledge was distilled from ResNet-50, which serves as the teacher model. Experimental results demonstrate significant enhancements and improvements in the student model, achieving a mean Average Precision (mAP) of 83.47%. Additionally, the model exhibits minimal computational expense, with only 5.7 million parameters. The proposed knowledge distillation framework successfully addresses the computational constraints associated with ViT deployment in agricultural robotics while preserving detection accuracy for weed detection applications.</p>","PeriodicalId":47597,"journal":{"name":"Frontiers in Robotics and AI","volume":"12 ","pages":"1654074"},"PeriodicalIF":3.0,"publicationDate":"2025-09-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12460097/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145187215","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}