Grasping Pose Detection for Loose Stacked Object Based on Convolutional Neural Network With Multiple Self-Powered Sensors Information

IEEE Sensors Journal, Год журнала: 2022, Номер 23(18), С. 20619 - 20632

Опубликована: Авг. 5, 2022

There are a variety of objects, random postures and multiple objects stacked in disorganized manner unstructured home applications, which leads to the object grasping posture estimation planning based on machine vision become very complicated. This paper proposes method cluttering pose detection convolutional neural network with self-powered sensors information. Firstly, search strategy for candidate poses 3D point cloud is proposed, single-channel image dataset representing this established by using Bigbird dataset. Secondly, ResNet constructed rank filter single channel captured images bit pose. It also compared three mainstream classification networks, Inception V2, VGG-A LetNet, perception analysis function execution developed under ROS. The effective manipulator scene scattered piles realized results position combined information sensors, other networks. In environment experiment show that superior average success rate ResNet, InceptionV2, VGGA LetNet networks 90.67%, 82.67%, 86.67% 87.33% respectively, verifies effectiveness superiority deep learn-based model proposed paper.

Язык: Английский

---

Photoelastic Stress Field Recovery Using Deep Convolutional Neural Network

Frontiers in Bioengineering and Biotechnology, Год журнала: 2022, Номер 10

Опубликована: Март 21, 2022

Recent work has shown that deep convolutional neural network is capable of solving inverse problems in computational imaging, and recovering the stress field loaded object from photoelastic fringe pattern can also be regarded as an problem process. However, formation affected by geometry specimen experimental configuration. When produces complex distribution, traditional analysis methods still face difficulty unwrapping. In this study, a based on encoder-decoder structure proposed, which accurately decode distribution information images generated under different configurations. The proposed method validated synthetic dataset, quality model evaluated using mean squared error (MSE), structural similarity index measure (SSIM), peak signal-to-noise ratio (PSNR), other evaluation indexes. results show recovery achieve average performance more than 0.99 SSIM.

Язык: Английский

---

Biomimetic Vision for Zoom Object Detection Based on Improved Vertical Grid Number YOLO Algorithm

Frontiers in Bioengineering and Biotechnology, Год журнала: 2022, Номер 10

Опубликована: Май 20, 2022

With the development of bionic computer vision for images processing, researchers have easily obtained high-resolution zoom sensing images. The drones equipped with high-definition cameras has greatly increased sample size and image segmentation target detection are important links during process information. As biomimetic remote usually prone to blur distortion in imaging, transmission processing stages, this paper improves vertical grid number YOLO algorithm. Firstly, light shade a were abstracted, grey-level cooccurrence matrix extracted feature parameters quantitatively describe texture characteristics image. Simple Linear Iterative Clustering (SLIC) superpixel method was used achieve light/dark scenes, saliency area obtained. Secondly, model segmenting dark scenes established made dataset meet recognition standard. Due refraction passing through lens other factors, difference contour boundary value between pixel background would make it difficult detect target, pixels main part separated be sharper edge detection. Thirdly, algorithm an improved proposed real time on processed array. adjusted aspect ratio modified grids network structure by using 20 convolutional layers five maximum aggregation layers, which more accurately adapted "short coarse" identified object information density. Finally, comparison mainstream algorithms different environments, test results aid showed that high spatial resolution images, higher accuracy than had real-time performance accuracy.

Язык: Английский

---

Performance evaluation of PSO-PID and PSO-FLC for continuum robot’s developed modeling and control

Scientific Reports, Год журнала: 2024, Номер 14(1)

Опубликована: Янв. 6, 2024

Continuum robots are complex structures that require sophisticated modeling and control methods to achieve accurate position motion tracking along desired trajectories. They highly coupled, nonlinear systems with multiple degrees of freedom pose a significant challenge for conventional approaches. In this paper, we propose system dynamic model based on the Euler-Lagrange formulation assumption piecewise constant curvature (PCC), where accounts elasticity gravity effects continuum robot. We also develop apply particle swarm optimization (PSO) algorithm optimize parameters our developed controllers: an inverse proportional integral derivative (PID) controller fuzzy logic (FLC), use time absolute error (ITAE) as objective function PSO algorithm. validate proposed optimized controllers through different designed trajectories, simulated using unique animated MATLAB simulation. The results show PSO-PID improves rise time, overshoot percentage, settling by 16.3%, 31.1%, 64.9%, respectively, compared PID without PSO. PSO-FLC shows best performance among all controllers, 0.7 s 0.4 s, leading highest level precision in trajectory tracking. ITAE is 11.4% 29.9% lower than FLC respectively.

Язык: Английский

---

Multi-Objective Location and Mapping Based on Deep Learning and Visual Slam

Sensors, Год журнала: 2022, Номер 22(19), С. 7576 - 7576

Опубликована: Окт. 6, 2022

Simultaneous localization and mapping (SLAM) technology can be used to locate build maps in unknown environments, but the constructed often suffer from poor readability interactivity, primary secondary information map cannot accurately grasped. For intelligent robots interact meaningful ways with their environment, they must understand both geometric semantic properties of scene surrounding them. Our proposed method not only reduce absolute positional errors (APE) improve positioning performance system also construct object-oriented dense point cloud output model each object reconstruct indoor scene. In fact, eight categories objects are for detection using coco weights our experiments, most actual reconstructed theory. Experiments show that number points is significantly reduced. The average error Technical University Munich (TUM) datasets very small. camera reduced introduction constraints, improved. At same time, algorithm segment environment high accuracy.

Язык: Английский

---

Kinematics characteristics analysis of a 3-UPS/S parallel airborne stabilized platform

Aerospace Science and Technology, Год журнала: 2023, Номер 134, С. 108163 - 108163

Опубликована: Фев. 3, 2023

Язык: Английский

---

Kinematics inverse solution of assembly robot based on improved particle swarm optimization

Robotica, Год журнала: 2024, Номер 42(3), С. 833 - 845

Опубликована: Янв. 4, 2024

Abstract Inverse kinematics of robot is the basis assembly, which directly determines pose robot. Because traditional inverse solution algorithm limited by topology structure, singular and accuracy, it affects use robots. In order to solve above problems, an improved particle swarm optimization (PSO) proposed problem This initializes population based on joint angle limitations, accelerating convergence speed algorithm. avoid falling into local optima premature convergence, we have a nonlinear weight strategy update position particles, enhancing algorithm’s search ability, in addition introducing penalty function eliminate particles exceeding limits. Finally, positions common points are selected PUMA 560 redundant for simulation verification. The results show that, compared with other algorithms, PSO has higher accuracy better solving solution, certain universality, provides new assembly

Язык: Английский

---

Construction Project Cost Prediction Method Based on Improved BiLSTM

Applied Sciences, Год журнала: 2024, Номер 14(3), С. 978 - 978

Опубликована: Янв. 23, 2024

In construction project management, accurate cost forecasting is critical for ensuring informed decision making. this article, a prediction method based on an improved bidirectional long- and short-term memory (BiLSTM) network proposed to address the high interactivity among data difficulty in feature extraction. Firstly, correlation between cost-influencing factors unilateral calculated via grey analysis select characteristic index. Secondly, BiLSTM used capture temporal interactions at deep level, hybrid attention mechanism incorporated enhance model’s extraction capability comprehensively features data. Finally, hyperparameter optimisation particle swarm algorithm using accuracy as fitness function of algorithm. The MAE, RMSE, MPE, MAPE, coefficient determination simulated results dataset are 7.487, 8.936, 0.236, 0.393, 0.996%, respectively, where MPE positive coefficient. This avoids serious consequences underestimating cost. Compared with unimproved BiLSTM, MAPE reduced by 15.271, 18.193, 0.784%, which reflects superiority effectiveness can provide technical support estimation field.

Язык: Английский

---

Target Detection Based on Two-Stream Convolution Neural Network With Self-Powered Sensors Information

IEEE Sensors Journal, Год журнала: 2022, Номер 23(18), С. 20681 - 20690

Опубликована: Ноя. 16, 2022

With the rapid development of artificial intelligence, a neural network is widely used in various fields. The target detection algorithm mainly based on network, but accuracy greatly related to complexity scene and texture. A RGB-D image from perspective lightweight model integration depth map overcome weak environmental illumination with self-powered sensors information proposed. This article analyzes structure YOLOv4 MobileNet, compares variation parameter numbers between depthwise separable convolution convolutional networks, combines advantages MobileNetv3 network. main three effective feature layers replaced by for initial layer extraction strengthen At same time, standard models are convolution. proposed method compared YOLOv4-MobileNetv3 this article, experimental results show that retains its original accuracy, size about 23% model, processing speed 42% higher than can still reach 83% environment poor lighting conditions.

Язык: Английский

---

Real-time visual SLAM based YOLO-Fastest for dynamic scenes

Measurement Science and Technology, Год журнала: 2024, Номер 35(5), С. 056305 - 056305

Опубликована: Фев. 5, 2024

Abstract Within the realm of autonomous robotic navigation, simultaneous localization and mapping (SLAM) serves as a critical perception technology, drawing heightened attention in contemporary research. The traditional SLAM systems perform well static environments, but real physical world, dynamic objects can destroy geometric constraints system, further limiting its practical application world. In this paper, robust RGB-D system is proposed to expand number points scene by combining with YOLO-Fastest ensure effectiveness model construction, then based on that, new thresholding designed differentiate features objection bounding box, which takes advantage double polyline residuals after reprojection filter feature points. addition, two Gaussian models are constructed segment moving box depth image achieve effect similar instance segmentation under premise ensuring computational speed. experiments conducted sequences provided TUM dataset evaluate performance method, results show that root mean squared error metric absolute trajectory algorithm paper has at least 80% improvement compared ORB-SLAM2. Higher robustness environments both high low DS-SLAM Dynaslam, effectively provide intelligent navigation for mobile robots.

Язык: Английский

---