Robust segmentation in adverse weather conditions is crucial for autonomous driving. However, these scenes struggle with recognition and make annotations expensive, resulting poor performance. As a result, the Segment Anything Model (SAM) was recently proposed to finely segment spatial structure of provide powerful prior information, thus showing great promise resolving problems. SAM cannot be applied directly different geographic scales non-semantic outputs. To address issues, we propose...

The accurate detection of relevant vehicles, pedestrians, and other targets on the road plays a crucial role in ensuring safety autonomous driving. In recent years, object detectors based Transformers or CNNs have achieved excellent performance fully supervised paradigm. However, when trained model is directly applied to unfamiliar scenes where training data testing different distributions statistically, model’s may decrease dramatically. To address this issue, unsupervised domain adaptive...

Semantic segmentation based on optical images can provide comprehensive scene information for intelligent vehicle systems, thus aiding in perception and decision making. However, under adverse weather conditions (such as fog), the performance of methods be compromised due to incomplete observations. Considering success domain adaptation recent years, we believe it is reasonable transfer knowledge from clear existing annotated datasets with fog. Technically, follow main workflow previous...

The mining belt conveyor is one of the most important modules in coal mine, whose safety always be threatened by foreign objects. Although traditional target detection methods achieve promising results various computer vision tasks, performance heavily depends on sufficient labelled data. However, real-world production scenario, it difficult to acquire huge number images with obtained datasets lacking capacity and diversity are not suitable for training supervised learning-based objects...

Abstract Aiming at the current bottleneck problem that restricts practical application of robot system engineering in transmission line live operation, order to improve on-line and off-loading efficiency working robots on lines level as a whole, this paper proposes design operation method autonomous mechanism double-arm based UAV assistance. Firstly, realization process up-and-down was analyzed, which structure auxiliary hook lifting winch were designed. Then, force analysis carried out...

Test-time adaptation (TTA) has been proven to effectively improve the adaptability of deep learning semantic segmentation models facing continuous changeable scenes. However, most existing TTA algorithms lack an explicit exploration domain gaps, especially those based on visual prompts. To address these issues, this paper proposes a self-prompt strategy uncertainty, guiding model continuously focus regions with high uncertainty (i.e., larger gap). Specifically, we still use Mean-Teacher...

Novel view synthesis using neural radiance fields (NeRFs) for remote sensing images is important various applications. Traditional methods often use implicit representations modeling, which have slow rendering speeds and cannot directly obtain the structure of 3D scene. Some studies introduced explicit representations, such as point clouds voxels, but this kind method produces holes when processing large-scale scenes from images. In addition, NeRFs with expression are more susceptible to...

In traditional 3D reconstruction using UAV images, only radiance information, which is treated as a geometric constraint, used in feature matching, allowing for the restoration of scene’s structure. After introducing supervision, NeRF can adjust geometry fixed-ray direction, resulting smaller search space and higher robustness. Considering lack construction methods aerial scenarios, we propose new point sampling method, generated imaging model, compatible with global geographic coordinate...

In the paper, a nine-index evaluation method of soil corrosive property to grounding grid is established based on analysis physical and chemical properties in 120 transformer substation test stations, inspection index corrosion condition about buried metals as well results influence temperature, water content, pH value, chlorine ion sulfate content other factors metal corrosion. The simple practical, comprehensive are considered, accuracy high, thereby providing scientific basis for...

Overhead transmission line energized maintenance is a necessary means to maintain the safe and stable operation of electric power, common tasks require manual tower operation, inspection robots are gradually becoming powerful tool as an alternative or auxiliary but conventional have problems such difficulty in getting on off line, single function, weak intelligent behavior, etc. Therefore, this paper proposes new method cooperative between large-load UAV power robot, single-arm robot...

Abstract With the rapid development of high-voltage transmission networks, coverage overhead lines has become wider and wider. The overhaul live work is also becoming more frequent. In order to improve efficiency maintenance, this paper proposes a mechanism that uses large-load unmanned aerial vehicle (UAV) carry wire-hanging by using combination U AV machine vision. new method based on Local Binary Pattern (LBP)vision system identify wire realizes one-key hookup, achieves purpose assisting...

The auxiliary material is often used to reduce the resistance of transmission line tower grounding grid for limitation extension and vertical reduction measures. For lack unified standard guidance when modules are in actual project, some problems occurred application such as blind construction low efficiency. In this paper, common lightning protection software CDEGS simulate calculate influencing factors module grid. First all, simulation carried out laying position density "one" shaped...

Smart grid gets more and popular today.Distributed generation is one of the key technologies, especially, integration problem distributed an important issue.Especially, location capacity play role for performance distribution network.In this paper, optimization model to minimize loss cost unsatisfied demand given.This based on a reliability computing method which avoiding power flow calculation in previous work.Then used IEEE-123 nodes experiment network result five placement got.

Abstract As an important development direction of substation equipment, a large number GIS equipment are installed in the power system. Once has internal faults, it is very inconvenient to find and deal with. This article designs miniature maintenance robot for which can be put into busbar through hand hole cover; ability walk tube; carries small camera device, check problems return images; The equipped with manipulator, vacuuming grasping functions, complete simple cleaning bolt tightening...

Abstract It has important theoretical significance and practical application value to improve the work efficiency reduce hazard of personnel through assistance robot for live transmission line instead manual work. However, in actual process robot, it takes a lot time, manpower material resources go online offline, overall is greatly reduced. In view above problems, this article proposes design scheme an automated offline system working robot. First, principle analyzed detail, basic...

Abstract Metal cutting tool is an important part of machining, and its performance directly affects the manufacturing efficiency machining quality products. With increasing demands in industry performance, efficiency, customization quick response, traditional design methods can no longer meet above requirements due to many repetitive work, large amount calculation, complex process low efficiency. Parametric has become a new development direction customized because fast, stable accurate...