Semantic segmentation of high-resolution remote sensing images is highly challenging due to the presence a complicated background, irregular target shapes, and similarities in appearance multiple categories. Most existing methods that rely only on simple fusion extracted multi-scale features often fail provide satisfactory results when there large difference sizes. Handling this problem through context extraction efficient features, paper we present an end-to-end adaptive feature network...

This paper shows how a robot arm can follow and grasp moving objects tracked by vision system, as is needed when human hands over an object to the during collaborative working. While being arbitrarily moved co-worker, set of likely grasps, generated learned planner, are evaluated online generate feasible with respect both: current configuration respecting target grasp; constraints finding collision-free trajectory reach that configuration. A task-based cost function enables relaxation...

We present early pilot-studies of a new international project, developing advanced robotics to handle nuclear waste. Despite enormous remote handling requirements, there has been remarkably little use robots by the industry. The few deployed have directly teleoperated in rudimentary ways, with no control methods or autonomy. Most is still done an aging workforce highly skilled experts, using 1960s style mechanical Master-Slave devices. In contrast, this paper explores how novice human...

Numerous grasp planning algorithms have been proposed since the 1980s. The grasping literature has expanded rapidly in recent years, building on greatly improved vision systems and computing power. Methods to plan stable grasps known objects (exact 3D model is available), familiar (e.g. exploiting a-priori for different of same category), or novel object shapes observed during task execution. Few these methods ever compared a systematic way, objective performance evaluation such complex...

This paper addresses the problem of grasping arbitrarily shaped objects, observed as partial point-clouds, without requiring: models physics parameters, training data, or other a-priori knowledge. A grasp metric is proposed based on Local Contact Moment (LoCoMo). LoCoMo combines zero-moment shift features, both hand and object surface patches, to determine local similarity. then used search for a set feasible poses with associated likelihoods. overcomes some limitations classical planners...

This paper addresses the problem of estimating configuration robots with no proprioceptive sensors and kinematic constraints while performing tasks. Our work is motivated by use unsensored (industrial) manipulators, currently teleoperated in rudimentary ways, hazardous environments, such as nuclear decommissioning. For robots, basic are often unavailable. Even if radiation-hardened could be retrofitted, manipulators typically deployed on a mobile base, equipped powerful end-effector tools...

Abstract 3D edge features, which represent the boundaries between different objects or surfaces in a scene, are crucial for many computer vision tasks, including object recognition, tracking, and segmentation. They also have numerous real-world applications field of robotics, such as vision-guided grasping manipulation objects. To extract these features noisy depth data, reliable detectors indispensable. However, currently available detection methods either highly parameterized require...

Polarimetric synthetic aperture radar (PolSAR) image classification is a recent technology with great practical value in the field of remote sensing. However, due to time-consuming and labor-intensive data collection, there are few labeled datasets available. Furthermore, most available state-of-the-art methods heavily suffer from speckle noise. To solve these problems, this paper, novel semi-supervised algorithm based on self-training superpixels proposed. First, Pauli-RGB over-segmented...

Object detection has always been a challenging task in the field of computer vision due to complex background, large scale variation and many small objects, which are especially pronounced for remote sensing imagery. In recent years, object with development deep learning also made great breakthroughs. At present, almost all state-of-the-art detectors rely on pre-defined anchor boxes However, too will introduce number hyper-parameters, not only increase memory footprint, but computational...

Summary As an imaging system, scanning electron microscope (SEM) performs important role in autonomous micro‐nanomanipulation applications. When it comes to the sub micrometer range and at high speeds, images produced by SEM are noisy need be evaluated or corrected beforehand. In this article, quality of a tungsten gun has been quantifying level image signal‐to‐noise ratio (SNR). order determine SNR, efficient online monitoring method is developed based on nonlinear filtering using single...

Positioning of micro-nanoobjects inside a scanning electron microscope (SEM) for manipulation is key and challenging task to perform. Often it performed by skilled operators via teleoperation, which tedious lacks repeatability. In this paper, rendering as an image-guided problem, we present frequency domain scheme automatic control positioning platform movements. The designed controller uses the relative global image motion computed using spectral information images visual signal can provide...

Automatically searching for optimal hyper parameters is of crucial importance applying machine learning algorithms in practice. However, there are concerns regarding the tradeoff between efficiency and effectiveness current approaches when faced with expensive function evaluations. In this paper, a novel efficient hyper-parameter optimization algorithm proposed (called MARSAOP), which multivariate spline functions used as surrogate dynamic coordinate search approach employed to generate...

Scanning Electron Microscope (SEM) image acquisition is mostly affected by the time varying motion of pixel positions in consecutive images, a phenomenon called drift. In order to perform accurate measurements using SEM, it necessary compensate this drift advance. Most existing compensation methods were developed correlation technique. paper, we present an registration-based method, where correction on distorted performed computing homography, keypoint correspondences between images. Four...

In this paper, we present a novel concept and primary investigations regarding automated unfastening of hexagonal nuts by means surface exploration with compliant robot. contrast to the conventional industrial approaches that rely on custom-designed motorised tools mechanical tool changers, propose use robot fingers position, grasp unfasten unknown random-sized nuts, which are arbitrarily positioned in robot’s task space. Inspired how visually impaired people handle objects, work,...

This paper presents two visual servoing approaches for nanopositioning in a scanning electron microscope (SEM). The first approach uses the total pixel intensities of an image as measurements designing control law. positioning error and platform are directly linked with intensity variations. second is frequency domain method that Fourier transform to compute relative motion between images. In this case, law designed minimize i.e. 2D current desired images by controlling movement. Both...

Depth estimation for micronanomanipulation inside a scanning electron microscope (SEM) is always major concern. So far, in the literature, various methods have been proposed based on stereoscopic imaging. Most of them require external hardware unit or manual interaction during process. In this paper, solely relying image sharpness information, we present new technique to estimate depth real time. To improve accuracy as well rapidity method, consider both autofocus and visual servoing...

This paper presents an assisted telemanipulation approach with integrated grasp planning. It also studies how the human teleoperation performance benefits from incorporated visual and haptic cues while manipulating objects in cluttered environments. The developed system combines widely used master-slave our previous model-free learning-free grasping algorithm by means of a dynamic re-ranking strategy semi-autonomous reach-to-grasptrajectory guidance. proposed metric helps dynamically...

To boost the circular economy of electric vehicle battery industry, an accurate assessment state health retired batteries is essential to assign them appropriate value in post automotive market and material degradation before recycling. In practice, advanced testing techniques are usually limited laboratory benches at cell level hardly used industrial environment module or pack level. This necessitates developing recycling facilities that can handle undertakings for many with different form...

Fast and reliable autofocusing methods are essential for performing automatic nano-objects positioning tasks using a scanning electron microscope (SEM). So far in the literature, various algorithms have been proposed utilizing sharpness measure to compute best focus. Most of them based on iterative search approaches; applying function over total range focus find an image in-focus. In this paper, new, fast direct method has presented idea traditional visual servoing control step adaptive...

Task-aware robotic grasping is critical if robots are to successfully cooperate with humans. The choice of a grasp multi-faceted; however, the task perform primes this in terms hand shaping and placement on object. This strategy particularly important for robot companion, as it can potentially hinder success collaboration In work, we investigate how different strategies passer influence performance perceptions interaction human receiver. Our findings suggest that accounts subsequent receiver...