Frontiers in Robotics and AI最新文献

{"title":"Task-specific CNN size reduction through content-specific pruning.","authors":"Nurbek Konyrbaev, Martin Lukac, Sabit Ibadulla, Askhat Diveev, Elena Sofronova, Asem Galymzhankyzy","doi":"10.3389/frobt.2025.1552068","DOIUrl":"10.3389/frobt.2025.1552068","url":null,"abstract":"<p><p>The widespread and growing use of flying unmanned aerial vehicles (UAVs) is attributed to their high spatial mobility, autonomous control, and lower cost compared to usual manned flying vehicles. Applications, such as surveying, searching, or scanning the environment with application-specific sensors, have made extensive use of UAVs in fields like agriculture, geography, forestry, and biology. However, due to the large number of applications and types of UAVs, limited power has to be taken into account when designing task-specific software for a target UAV. In particular, the power constraints of smaller UAVs will generally necessitate reducing power consumption by limiting functionality, decreasing their movement radius, or increasing their level of autonomy. Reducing the overhead of control and decision-making software onboard is one approach to increasing the autonomy of UAVs. Specifically, we can make the onboard control software more efficient and focused on specific tasks, which means it will need less computing power than a general-purpose algorithm. In this work, we focus on reducing the size of the computer vision object classification algorithm. We define different tasks by specifying which objects the UAV must recognize, and we construct a convolutional neural network (CNN) for each specific classification. However, rather than creating a custom CNN that requires its dataset, we begin with a pre-trained general-purpose classifier. We then choose specific groups of objects to recognize, and by using response-based pruning (RBP), we simplify the general-purpose CNN to fit our specific needs. We evaluate the pruned models in various scenarios. The results indicate that the evaluated task-specific pruning can reduce the size of the neural model and increase the accuracy of the classification tasks. For small UAVs intended for tasks with reduced visual content, the proposed method solves both the size reduction and individual model training problems.</p>","PeriodicalId":47597,"journal":{"name":"Frontiers in Robotics and AI","volume":"12 ","pages":"1552068"},"PeriodicalIF":3.0,"publicationDate":"2025-09-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12460118/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145187210","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":"Avoidance behaviours of farmed Atlantic salmon (<i>Salmo salar</i> L.) to artificial sound and light: a case study of net-pen mariculture in Norway.","authors":"Qin Zhang, Nina Bloecher, Linn Danielsen Evjemo, Martin Føre, Eleni Kelasidi","doi":"10.3389/frobt.2025.1657567","DOIUrl":"10.3389/frobt.2025.1657567","url":null,"abstract":"<p><p>Intensive finfish aquaculture is increasingly relying on enabling technologies and solutions such as sensor systems, robotics, and other machinery. Together with conventional farming equipment, these systems may emanate acoustic noise and artificial light, impacting the pen environment. Farmed fish have been observed to respond behaviourally and/or physiologically to anthropogenic sounds and lights, indicating a stress reaction that could impair welfare and health. This study aimed to investigate how farmed Atlantic salmon respond to such stimuli, with direct implications for the design and operation of robotic and mechanised systems in sea pens. We conducted experiments where we systematically exposed adult farmed Atlantic salmon in commercial net pens to sounds of frequencies within the range common to farm equipment (100-1,000 Hz), and submerged lights at 8 and 12 m with four different intensities (600 lx-14,500 lx). Data was analysed using sonar data and a deep learning (DL) based method for processing that automatically identified fish distribution patterns and estimated the average avoidance distance to the sound/light source. The fish fled from the sound source while playing sounds of 400 Hz, while sounds at other frequencies did not elicit a response. The response to light intensity depended on deployment depth, with the fish moving closer to the source when intensity was increased at 8 m depth, but conversely moving further away with increasing density when it was placed at 12 m. These outcomes are important inputs for the design of equipment, autonomous vehicles, robotic interventions and operations at commercial farms to ensure that their sound and light emissions have minimal impact on the fish, thereby reducing the potential of induced stress.</p>","PeriodicalId":47597,"journal":{"name":"Frontiers in Robotics and AI","volume":"12 ","pages":"1657567"},"PeriodicalIF":3.0,"publicationDate":"2025-09-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12460098/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145187194","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":"The impact of hotel robots' service quality on continuance intention: the moderating effect of personal innovation.","authors":"Yaru Shi, Weifang Zhan, Chengpeng Lin","doi":"10.3389/frobt.2025.1667123","DOIUrl":"10.3389/frobt.2025.1667123","url":null,"abstract":"<p><p>As service robots are increasingly integrated into the hotel industry to enhance operational efficiency and customer experience, understanding consumers' responses to robotic services has become a critical research agenda. However, empirical evidence on how customers evaluate the service quality of hotel robots and how these evaluations influence their continuance intention remains limited. Drawing on the SERVQUAL framework, this study redefines service quality in the context of AI-powered hotel robots through five dimensions: reliability, assurance, entertainment, anthropomorphism, and tangibles. Furthermore, the study explores the moderating role of personal innovativeness in the relationship between perceived service quality and continuance intention. Data were collected via an online survey from 400 Generation Z consumers in China who had prior experience with item-delivery robots in hotel settings. The results indicate that assurance, entertainment, anthropomorphism, and tangibles have significant positive effects on continuance intention, while reliability does not show a statistically significant impact. Moreover, personal innovativeness significantly moderates the effects of certain service quality dimensions, suggesting that individual differences in technology readiness shape consumer reactions to robotic services. This study contributes to the literature by extending traditional service quality theory into the domain of human-robot interaction and by highlighting the nuanced mechanisms through which robot-specific service attributes influence user loyalty. Practical implications are offered for hotel managers seeking to optimize robot deployment strategies and improve guest engagement in technology-enhanced service environments.</p>","PeriodicalId":47597,"journal":{"name":"Frontiers in Robotics and AI","volume":"12 ","pages":"1667123"},"PeriodicalIF":3.0,"publicationDate":"2025-09-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12457782/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145151331","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":"Encouraging classroom activities for children using avatar robots: a field trial.","authors":"Megumi Kawata, Masashi Maeda, Hirokazu Kumazaki, Hiroko Kamide, Jun Baba, Naomi Matsuura, Hiroshi Ishiguro, Yuichiro Yoshikawa","doi":"10.3389/frobt.2025.1571804","DOIUrl":"10.3389/frobt.2025.1571804","url":null,"abstract":"<p><p>Educational institutions are facing a critical shortage of teachers worldwide. Consequently, the trend of introducing interactive robots into educational sites is growing. However, most previous research focused on specific subjects or time slots, and only a few studies have introduced interactive robots to participate in whole classroom activities with children routinely. This study investigates the use of avatar robots operated by multiple remote operators in elementary school classrooms. Over nine days, a 5th-grade class was observed to assess the robot's impact on student engagement, motivation, and peer interactions, and compared to classes where any avatar robots were not introduced. Key findings include improved student confidence in presentations, enhanced social interactions during recess, and positive feedback on the robot's role in supporting classroom activities. The results suggest that avatar robots, with consistent remote operation, can provide valuable educational support without strong negative reactions from students.</p>","PeriodicalId":47597,"journal":{"name":"Frontiers in Robotics and AI","volume":"12 ","pages":"1571804"},"PeriodicalIF":3.0,"publicationDate":"2025-09-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12454100/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145139115","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":"Developing reliability centered maintenance in automotive robotic welding machines for a tier 1 supplier.","authors":"H T O Alaka, K Mpofu, B I Ramatsetse, T A Adegbola, M O Adeoti","doi":"10.3389/frobt.2025.1620370","DOIUrl":"https://doi.org/10.3389/frobt.2025.1620370","url":null,"abstract":"<p><p>The study highlights the effectiveness of FMEA in robotic spot-welding operations, providing a systematic approach to enhancing performance in an automotive assembly line. Robotic welding industries depend on mechanized, programmable tools to automate welding processes, ensuring efficiency, reliability, and effective material handling. In the automotive sector, Tier 1 suppliers utilize robotic welding machines to produce high volumes of welded assemblies, with daily output exceeding 450 units. However, frequent equipment downtime due to maintenance challenges disrupts productivity and impacts customer satisfaction. This study aims to develop a Reliability-Centered Maintenance (RCM) approach for robotic welding industries, optimizing machine uptime, enhancing product quality, and reducing financial losses caused by unexpected failures. A 3-year dataset was analysed to identify the primary causes of downtime and their associated costs. Failure Modes and Effects Analysis (FMEA) was applied to assess failure modes, determine root causes, and calculate Risk Priority Numbers (RPNs), thereby formulating corrective actions to mitigate recurring failures and enhance operational efficiency. Findings revealed that maintenance-related issues accounted for 79% of total downtime, resulting in financial losses of R2,281,508.82 over 3 years. The application of FMEA provided a structured framework for prioritizing critical failure modes and implementing targeted corrective measures to reduce downtime and enhance overall reliability. To sustain high productivity and quality, it is recommended that robotic welding industries adopt proactive maintenance strategies based on FMEA findings. Regular monitoring, predictive maintenance, and workforce training will help minimize machine failures and optimize operational efficiency.</p>","PeriodicalId":47597,"journal":{"name":"Frontiers in Robotics and AI","volume":"12 ","pages":"1620370"},"PeriodicalIF":3.0,"publicationDate":"2025-09-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12454091/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145139072","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":"Diffusion models for robotic manipulation: a survey.","authors":"Rosa Wolf, Yitian Shi, Sheng Liu, Rania Rayyes","doi":"10.3389/frobt.2025.1606247","DOIUrl":"10.3389/frobt.2025.1606247","url":null,"abstract":"<p><p>Diffusion generative models have demonstrated remarkable success in visual domains such as image and video generation. They have also recently emerged as a promising approach in robotics, especially in robot manipulations. Diffusion models leverage a probabilistic framework, and they stand out with their ability to model multi-modal distributions and their robustness to high-dimensional input and output spaces. This survey provides a comprehensive review of state-of-the-art diffusion models in robotic manipulation, including grasp learning, trajectory planning, and data augmentation. Diffusion models for scene and image augmentation lie at the intersection of robotics and computer vision for vision-based tasks to enhance generalizability and data scarcity. This paper also presents the two main frameworks of diffusion models and their integration with imitation learning and reinforcement learning. In addition, it discusses the common architectures and benchmarks and points out the challenges and advantages of current state-of-the-art diffusion-based methods.</p>","PeriodicalId":47597,"journal":{"name":"Frontiers in Robotics and AI","volume":"12 ","pages":"1606247"},"PeriodicalIF":3.0,"publicationDate":"2025-09-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12454101/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145139173","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":"Editorial: Advancements in vibration control for space manipulators: actuators, algorithms, and material innovations.","authors":"Javad Tayebi, Ti Chen, Xiaofeng Wu, Anand Kumar Mishra","doi":"10.3389/frobt.2025.1681168","DOIUrl":"10.3389/frobt.2025.1681168","url":null,"abstract":"","PeriodicalId":47597,"journal":{"name":"Frontiers in Robotics and AI","volume":"12 ","pages":"1681168"},"PeriodicalIF":3.0,"publicationDate":"2025-09-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12446049/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145114457","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":"Combining vision and range sensors for AMCL localization in corridor environments with rectangular signs.","authors":"Paloma de la Puente, Germán Vega-Martínez, Patricia Javierre, Javier Laserna, Elena Martin-Arias","doi":"10.3389/frobt.2025.1652251","DOIUrl":"10.3389/frobt.2025.1652251","url":null,"abstract":"<p><p>Localization is widely recognized as a fundamental problem in mobile robotics. Even though robust localization methods do exist for many applications, it is difficult for them to succeed in complex environments and challenging situations. In particular, corridor-like environments present important issues for traditional range-based methods. The main contribution of this paper is the integration of new observation models into the popular AMCL ROS node, considering visual features obtained from the detection of rectangular landmarks. Visual rectangles are distinctive elements which are very common in man-made environments and should be detected and recognized in a robust manner. This hybrid approach is developed and evaluated both for the combination of an omnidirectional camera and a laser sensor (using artificial markers) and for RGB-D sensors (using natural rectangular features). For the latter, this work also introduces RIDGE, a novel algorithm for detecting projected quadrilaterals representing rectangles in images. Simulations and real world experiments are presented for both cases. As shown and discussed in the article, the proposed approach provides significant advantages for specific conditions and common scenarios such as long straight corridors.</p>","PeriodicalId":47597,"journal":{"name":"Frontiers in Robotics and AI","volume":"12 ","pages":"1652251"},"PeriodicalIF":3.0,"publicationDate":"2025-09-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12447077/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145114438","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":"Integrating emotional intelligence, memory architecture, and gestures to achieve empathetic humanoid robot interaction in an educational setting.","authors":"Fuze Sun, Lingyu Li, Shixiangyue Meng, Xiaoming Teng, Terry R Payne, Paul Craig","doi":"10.3389/frobt.2025.1635419","DOIUrl":"10.3389/frobt.2025.1635419","url":null,"abstract":"<p><p>This study investigates the integration of individual human traits into an empathetically adaptive educational robot tutor system designed to improve student engagement and learning outcomes with corresponding Engagement Vector measurements. While prior research in the field of Human-Robot Interaction (HRI) has examined the integration of the traits, such as emotional intelligence, memory-driven personalization, and non-verbal communication, by themselves, they have thus-far neglected to consider their synchronized integration into a cohesive, operational education framework. To address this gap, we customize a Multi-Modal Large Language Model (Llama 3.2 from Meta) deployed with modules for human-like traits (emotion, memory and gestures) into an AI-Agent framework. This constitutes the robot's intelligent core that mimics the human emotional system, memory architecture and gesture controller to allow the robot to behave more empathetically while recognizing and responding appropriately to the student's emotional state. It can also recall the student's past learning record and adapt its style of interaction accordingly. This allows the robot tutor to react to the student in a more sympathetic manner by delivering personalized verbal feedback synchronized with relevant gestures. Our study suggests the extent of this effect through the introduction of Engagement Vector Model which can be a benchmark for judging the quality of HRI experience. Quantitative and qualitative results demonstrate that such an empathetic responsive approach significantly improves student engagement and learning outcomes compared with a baseline humanoid robot without these human-like traits. This indicates that robot tutors with empathetic capabilities can create a more supportive, interactive learning experience that ultimately leads to better outcomes for the student.</p>","PeriodicalId":47597,"journal":{"name":"Frontiers in Robotics and AI","volume":"12 ","pages":"1635419"},"PeriodicalIF":3.0,"publicationDate":"2025-09-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12444663/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145114466","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":"Integrating large language models for intuitive robot navigation.","authors":"Ziheng Xue, Arturs Elksnis, Ning Wang","doi":"10.3389/frobt.2025.1627937","DOIUrl":"10.3389/frobt.2025.1627937","url":null,"abstract":"<p><p>Home assistance robots face challenges in natural language interaction, object detection, and navigation, mainly when operating in resource-constrained home environments, which limits their practical deployment. In this study, we propose an AI agent framework based on Large Language Models (LLMs), which includes EnvNet, RoutePlanner, and AIBrain, to explore solutions for these issues. Utilizing quantized LLMs allows the system to operate on resource-limited devices while maintaining robust interaction capabilities. Our proposed method shows promising results in improving natural language understanding and navigation accuracy in home environments, also providing a valuable exploration for deploying home assistance robots.</p>","PeriodicalId":47597,"journal":{"name":"Frontiers in Robotics and AI","volume":"12 ","pages":"1627937"},"PeriodicalIF":3.0,"publicationDate":"2025-09-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12444764/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145114503","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}