In this paper, a neural network (NN) controller is designed to suppress the vibration of flexible robotic manipulator system with input deadzone. The NN aims approximate unknown dynamics and eliminate effects deadzone in actuators. order describe more accurately, model constructed based on lumping spring-mass method. Full state feedback control proposed first output high-gain observer then devised make scheme practical. effect approximated by radial basis function (RBFNN) another RBFNN. able...

Adaptive neural networks (NNs) are employed for control design to suppress vibrations of a flexible robotic manipulator. To improve the accuracy in describing elastic deflection manipulator, system is modeled via lumped spring-mass approach. Full-state feedback as well output proposed separately. Aiming at achieving objective, uniform ultimate boundedness closed-loop ensured. Numerical simulations model carried out verify performance NN control. Finally, experiments given further validate...

A system that allows a robot to acquire model of its environment and use this navigate is described. The maps the as set snapshots world taken at target locations. uses an image-based local homing algorithm between neighboring Features approach include imaging acquires compact, 360 degrees representation image-based, qualitative without explicitly inferring three-dimensional structure from image. results experiment in typical indoor are described, argued navigation feasible alternative...

To promote operational intelligence, improve surface quality, and reduce manpower dependence, a novel high-bandwidth end-effector with active force control for robotic polishing was proposed. Using this as mini robot, macro-mini robot processing constructed, in which the macro provides posture during operations, whereas mini-robot constant control. By minimizing inertia along spindle configuration, obtains bandwidth of 200 Hz. Through series comparative experiments different contact forces...

Few-shot learning aims to correctly recognize query samples from unseen classes given a limited number of support samples, often by relying on global embeddings images. In this paper, we propose equip the backbone network with an attention agent, which is trained reinforcement learning. The policy gradient algorithm employed train agent towards adaptively localizing representative regions feature maps over time. We further design reward function based prediction held-out data, thus helping...

The existence of two- and three-finger grasps in the presence arbitrarily small friction is shown for 2D 3D smooth objects using a simple technique. No convexity assumed. finger gaits rotating planar object three four fingers proved. Additional results convex which describe two different are presented. purpose to move one or more at any time same maintain grip with others, thereby further than other ways it could be moved fixed grip.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML"...

A system that allows a robot to use model of its environment navigate is reported. The maps the as set snapshots world taken at target locations. uses an image-based local homing algorithm between neighboring and approach has imaging acquires compact, 360 degrees representation intensity waveform, image-based, qualitative without explicitly inferring three-dimensional structure from image. result experiment in typical indoor are They show navigation feasible alternative approaches using 3-D...

Abstract In this paper, we extend Open-set Semantic Segmentation (OSS) into a new image segmentation task called Generalized (GOSS). Previously, with well-known OSS, the intelligent agents only detect unknown regions without further processing, limiting their perception capacity of environment. It stands to reason that analysis detected pixels would be beneficial for agents’ decision-making. Therefore, propose GOSS, which holistically unifies abilities two well-defined tasks, i.e. OSS and...

A system able to control the Utah/MIT hand with VPL DataGlove has been developed. To get actual joint angles from sensor values, a least-squares fit is used find best-fit exponential curve for each sensor, and then correlation between sensors reduced by iterative elimination procedure. The calibration depends both on wearer particular being used. first level simple can be done in under 15 minutes little experience. second fixed requires no adjustments. hand, mapping applied, making...

Audio-visual zero-shot learning aims to classify samples consisting of a pair corresponding audio and video sequences from classes that are not present during training. An analysis the audio-visual data reveals large degree hyperbolicity, indicating potential benefit using hyperbolic transformation achieve curvature-aware geometric learning, with aim exploring more complex hierarchical structures for this task. The proposed approach employs novel loss function incorporates cross-modality...

Over past two decades, steady-state evoked potentials (SSVEP)-based brain computer interface (BCI) systems have been extensively developed. As we all know, signal processing algorithms play an important role in this BCI. However, there is no comprehensive review of the latest development for SSVEP-based By analyzing papers published authoritative journals nearly five years, preprocessing, feature extraction and classification modules are discussed detail. In addition, other aspects existed...

In order to improve the detection accuracy of objects at different scales, most recent studies applied multilayer architecture. However, extracted low-level feature in shallow layers may not work perfectly on performance due its less semantic information, especially for small objects. this paper, we propose a refined feature-fusion structure be integrated with single shot detector (SSD). To obtain rich representation ability mapping, fusion block, deconvolution operation is basically fuse...

Jet fuel leaks not only waste resources and increase costs but also pose a risk of emergency landings aviation accidents. With the blossom implementation deep learning, crack segmentation techniques have been rapidly developed in many fields. However, it is struggle to get accurate complete results because images' complex background environment. To address this issue, we collected labeled dataset 2824 images from surface aircraft tank, named CAFT2800. And article presents an atrous spatial...

The macro-mini robot polishing system has been widely used in processes where an auxiliary actuator is to control the force. Generally, this approach, moves motor, spindle and head which are considerable loads. Hence, if these loads carried by actuator, its holding force needs be increased consequently physical size inertia should increased. In order address problem, authors propose a novel mechanical design of mini reduce moving actuator. Experimental results have verified that proposed...

Although electroencephalography (EEG) brain-computer interface (BCI) has been quite successful, multi-command control is still one of the key issues for external applications. Multimodal BCI represents direction dealing with this problem. In our study, five healthy subjects performed exp eriment cooperatively. EEG and electromyography (EMG) were recorded synchronously. For individual EEG, after Laplacian filtering, C3 C4 channels determined. Then, was decomposed into third layer by wavelet...

The aim of BCI is to translate brain activity into a command for computer. For this purpose, the EEG signal processing plays an important role, especially in feature extraction. In paper, combined extraction method proposed left-right hand motor imagery BCI. power sensorimotor rhythm band and statistical features wavelet coefficients are used extracting support vector machine adopted pattern recognition imagery. performance tested by signals subject b g from datasets1 Competition IV. results...

Audio-visual generalized zero-shot learning (AV-GZSL) for video classification is a task where the model learns to identify unseen classes from multimodal audio-visual inputs. This combination of two equally challenging tasks one tries perform different modalities inputs, and other pushes achieve this in setting. The natural alignment between audio visual key addressing relatively unexplored task. predominant approach AV-GZSL has been learn better cross-modal attention input domains leverage...

In this paper, the interaction between human awareness design and epidemic dynamics over arbitrary directed networks has been studied. People gain fear of disease from three information sources a novel upon two infectious states is proposed to reduce susceptibility individuals, which in turn could affect threshold outbreak spreading system. The response applied continuous mean-field SEIV model illustrate effectiveness system global exponential stability analysis embedded with are paper....

This paper focuses on the control design and stability analysis of a Susceptible-Exposed-Infected-Vigilant (SEIV) epidemic model via adaptive complex networks. The network is designed empirically as state-dependent network, where structure keeps changing to inhibit propagation. recovery rate disease prevention are chosen scheme in system, both which closely associated with medical resources allocation. People may cut connection an infected neighbor reduce frequency go out when occurs. In...

In order to improve the robot's perception ability in complicated environment, especially unstructured a framework of 3D object detection and pose estimation using single shot detector (SSD) modified LineMOD template matching is proposed, which can detect multiple objects estimate their simultaneously. Firstly, initial (the first detection) realized by network therefore region interest (RoI) target are generated. then applied provide candidate templates. These calculated templates grouped...

In this paper, we employ the adaptive Neural Network (NN) control to supress vibration a flexible robotic manipulator with proper tip-payload. The model of system is presented Lumped Spring-Mass method which can enhance accuracy in reflecting elastic vibrations original system. Full-state feedback carried out approximate unmeasured variables paper. Uniform ultimate boundedness (UUB) achieved via Lyapunov's direct theory. Simulations for lumped are verify effectiveness proposed NN strategies,...