Rapid development of artificial intelligence motivates researchers to expand the capabilities of intelligent and autonomous robots. In many robotic applications, robots are required to make planning decisions based on...
详细信息
Rapid development of artificial intelligence motivates researchers to expand the capabilities of intelligent and autonomous robots. In many robotic applications, robots are required to make planning decisions based on perceptual information to achieve diverse goals in an efficient and effective way. The planning problem has been investigated in active robot vision, in which a robot analyzes its environment and its own state in order to move sensors to obtain more useful information under certain constraints. View planning,which aims to find the best view sequence for a sensor,is one of the most challenging issues in active robot vision. The quality and efficiency of view planning are critical for many robot systems and are influenced by the nature of their tasks, hardware conditions, scanning states, and planning strategies. In this paper, wefirst summarize some basic concepts of active robot vision, and then review representative work on systems,algorithms and applications from four perspectives:object reconstruction, scene reconstruction, object recognition, and pose ***, some potential directions are outlined for future work.
vision tracking is a key component of a video sequence. It is the process of locating single or multiple moving objects over time using one or many cameras. The latter's function consists of detecting, categorizin...
详细信息
vision tracking is a key component of a video sequence. It is the process of locating single or multiple moving objects over time using one or many cameras. The latter's function consists of detecting, categorizing, and tracking. The development of the trustworthy solution for video sequence analysis opens up new horizons for a variety of applications, including intelligent transportation systems, biomedical, agriculture, human-machine interaction, augmented reality, video surveillance, robots, and many crucial research areas. To make efficient models, there are challenges in video observation to deal with, such as problems with the environment, light variation, pose variation, motion blur, clutter, occlusion, and so on. In this paper, we present several techniques that addressed the issues of detecting and tracking multiple targets on video sequences. The proposed comparative study relied on different methodologies. This paper's purpose is to list various approaches, classify them, and compare them, using the Weighted Scoring Model (WSM) comparison method. This includes studying these algorithms, selecting relevant comparison criteria, assigning weights for each criterion, and lastly computing scores. The obtained results of this study will reveal the strong and weak points of each algorithm mentioned and discussed.
Human-Robot Collaboration (HRC) enabling mechanisms require real-time detection of potential collisions among human and robots. Taking under consideration the already existing standards and the literature, most of col...
详细信息
This work focuses on the IoT Robotic Things and machine vision and perception algorithms to improve smart manufacturing in Industry 4.0. This paper exemplifies potential enhancements of IoRT systems by utilizing deep ...
详细信息
ISBN:
(数字)9798331518363
ISBN:
(纸本)9798331518370
This work focuses on the IoT Robotic Things and machine vision and perception algorithms to improve smart manufacturing in Industry 4.0. This paper exemplifies potential enhancements of IoRT systems by utilizing deep learning and sensor fusion strategies. Hence, through Convolutional Neural Networks (CNNs) for detecting and classifying objects and through image segmentation such as U-Net, the system analyzes large, dense and rich visual data typical of industrial scenarios. Pose estimation and 3D reconstruction are employed to develop highly refined spatial perceptive abilities. Data from multiple sensors is fused using Sensor fusion techniques such as Kalman and Particle Filters; thus improving the situational awareness, for closed loop decision making. Since the analysis happens at the edge side, it happens in real-time with very little latency. Both data augmentation and synthetic data enhance the stability and, in turn, the robustness of the model. This approach not only improves manufacturing processes to increase precision and efficiency but also sets a practical base for future intelligent and self-organized technical systems making IoRT a part of smart manufacturing.
Robotic manipulation holds the potential to replace humans in the execution of tedious or dangerous tasks. However, control-based approaches are not suitable due to the difficulty of formally describing open-world man...
详细信息
ISBN:
(数字)9798350377705
ISBN:
(纸本)9798350377712
Robotic manipulation holds the potential to replace humans in the execution of tedious or dangerous tasks. However, control-based approaches are not suitable due to the difficulty of formally describing open-world manipulation in reality, and the inefficiency of existing learning methods. Therefore, applying manipulation in a wide range of scenarios presents significant challenges. In this study, we propose a novel framework for skill learning in robotic manipulation called Tactile active Inference Reinforcement Learning (TactileAIRL), aimed at achieving efficient learning. To enhance the performance of reinforcement learning (RL), we introduce active inference, which integrates model-based techniques and intrinsic curiosity into the RL process. This integration improves the algorithm’s training efficiency and adaptability to sparse rewards. Additionally, we have designed universal tactile static and dynamic features based on vision-based tactile sensors, making our framework scalable to many manipulation tasks learning involving tactile feedback. Simulation results demonstrate that our method achieves significantly high training efficiency in objects pushing tasks. It enables agents to excel in both dense and sparse reward tasks with just few interaction episodes, surpassing the SAC baseline. Furthermore, we conduct physical experiments on a gripper screwing task using our method, which showcases the algorithm’s rapid learning capability and its potential for practical applications.
As deep neural networks are spreading to almost all fields, flight systems in the unmanned aerial vehicle (UAV) domain are undergoing various transitions to intelligent systems. Among these transitions-in a bid to red...
详细信息
As deep neural networks are spreading to almost all fields, flight systems in the unmanned aerial vehicle (UAV) domain are undergoing various transitions to intelligent systems. Among these transitions-in a bid to reduce flight risk-is the active research domain of autonomous navigation for intelligent UAVs. The autonomous trail-following flight system that this letter introduces can safely consolidate flight control and mission control within the latest commercial hardware platform. The resource usage and degradation of pass-through delay in vision-based convolutional neural network workloads show that virtualisation overhead is not significantly negative, and the overall performance of the introduced system is acceptable. Real-time cooperation is also verified as achievable-in that the workloads incur minimal communication delay-between the controls. Finally, the actual field test analysis demonstrates the applicability of our autonomous UAV system, whereby our system controls the UAV to follow the centre of a set trail.
The field of traditional mobile robot navigation has undergone a gradual transformation, evolving into a standardized and procedural research domain. Through a fresh cognitive perspective on this navigation process, a...
The field of traditional mobile robot navigation has undergone a gradual transformation, evolving into a standardized and procedural research domain. Through a fresh cognitive perspective on this navigation process, a bridge emerges between robotic researchers and cognitive neuroscientists, thereby fostering the growth of interdisciplinary exploration. This article meticulously examines three pivotal phases of robot navigation: information acquisition, simultaneous localization and mapping (SLAM), and path planning. These phases are intricately linked to pertinent cognitive research. Furthermore, a comprehensive survey of existing biomimetic and neuromorphic algorithms is presented, highlighting their applications in augmenting robot navigation capabilities. Ultimately, this article charts a visionary course for the future of robot navigation, envisioning its potential in facilitating more intelligent and efficient navigation techniques.
With the advent of Industry 5.0 and the rise of human-centered intelligent manufacturing, people have paid increasing attention to the issue of security in human-machine collaboration. Developing safe human-robot coop...
详细信息
ISBN:
(数字)9798350363173
ISBN:
(纸本)9798350363180
With the advent of Industry 5.0 and the rise of human-centered intelligent manufacturing, people have paid increasing attention to the issue of security in human-machine collaboration. Developing safe human-robot cooperation in constrained environments has emerged as the primary area of research interest. vision systems with deep learning have gradually supplanted more conventional approaches, such electronic wearables, electronic fences, and lidar techniques, to ensure safe collaboration. Object identification and posture estimation are two techniques that are currently in use to forecast distances in space more accurately. These techniques can track the approximate locations of humans and robots in real-time, significantly lowering the likelihood of safety incidents. Still, more accurate evaluation of the relative positions of humans and robots is needed for effective collaboration. This paper suggests SCC-HRNet, an efficient key point recognition technique. SCC-HRNet is able to find the important feature points of both humans and robots more precisely in dual-camera human-robot safe collaboration scenarios. Using our human-robot collaboration dataset, SCC-HRNet outperforms other algorithms with an average precision gain of 1.6%, correctly identifying key points.
The paper discusses the problem of creating an autonomous SCARA robot that can precisely carry out pick-and-place operations in unstructured settings. The suggested method combines computervision and deep learning al...
详细信息
ISBN:
(数字)9798331527549
ISBN:
(纸本)9798331527556
The paper discusses the problem of creating an autonomous SCARA robot that can precisely carry out pick-and-place operations in unstructured settings. The suggested method combines computervision and deep learning algorithms to provide accurate 6D pose estimation and reliable object detection. Combining YOLOv8 for object detection, hand-eye calibration for precise transformation to robot coordinates, and Principal Component Analysis (PCA) for orientation estimation within the generated point cloud, a novel method for estimating object pose is presented. Monocular depth estimation techniques are used to extract depth information from RGB images to create point clouds. The results demonstrate the potential of combining advanced algorithms with user-friendly design for robotics automation, showing notable gains in pose estimation accuracy and task execution efficiency.
Fully autonomous mobile robots have the potential to revolutionize various industries, from warehouse management to hospital logistics and last-mile deliveries. However, a significant obstacle to achieving reliable au...
Fully autonomous mobile robots have the potential to revolutionize various industries, from warehouse management to hospital logistics and last-mile deliveries. However, a significant obstacle to achieving reliable autonomy lies in the high computational and energy requirements. In response to this challenge, our paper introduces two innovative algorithms: the Pure Image Segmentation Approach (PISA) and the UNet Based Approach to Semantic Segmentation (UBASS). PISA leverages classical computervision techniques, offering a fresh perspective on solving crucial tasks such as object detection, object avoidance, and lane detection. In contrast, UBASS harnesses the power of deep learning algorithms for semantic segmentation, unlocking new capabilities in robot perception. Our experiments showcase the effectiveness of these algorithms, demonstrating their accuracy and computational efficiency. Notably, PISA and UBASS outperform or match traditional techniques, including End-to-End Deep Learning and Canny Edge Detection, in terms of both task performance and resource utilization. This research contributes to the advancement of autonomous mobile robotics by offering practical and efficient solutions for navigation and perception challenges. By combining classic and contemporary approaches, we aim to inspire further research in the field, ultimately paving the way for more accessible and dependable autonomous mobile robots.
暂无评论