Road information is a kind of important geographic information. extracted from remote sensing images widely used in map, traffic, navigation, military and many other fields. However, the autonomous extraction road high resolution has some problems such as incoherence, incompleteness poor connectivity, therefore, semantic segmentation model for roads images, called AGD-Linknet, proposed, which integrates attention mechanisms, gated decoder block, dilated convolution. This mainly consists...

Genetic algorithm is a kind of way to solve complex problems effectively, for it not bound by the restrictive assumptions search space, and doesn't require assumption conditions such as continuity derivatives. So this has its advantage that traditional can compared. uses multi-point search. In each iteration, new individuals are generated mating mutation, so searching range be expanded, local optimal solution effectively prevented. paper, real number encoding method based change course...

This paper proposes an efficient wayfinding strategy for robot navigation in regionalized environments by designing a spatial knowledge model (RSK model) and region-based algorithm, i.e., fine-to-coarse A* (FTC-A*) search algorithm. First, the RSK model, which imitates representation of human brain, is presented to describe environments. The that are divided into regions represented hierarchical nested structure where small grouped together form superordinate regions. Second, on basis FTC-A*...

In this paper, we investigate the formation control of multiple mobile robots in an unknown environment with obstacles. A hybrid approach based on leader-follower scheme is proposed for control. The approach, using relative position bias between virtual targets and follower robots, includes active behavior-based strategy target structure. Furthermore, a supervision mechanism designed to ensure integrity. Real-world experiments are performed both obstacle avoidance, three nonholonomic results...

Recently, rapidly-exploring random trees(RRT) is widely used in path planning for its nature of single-query. The optimized algorithm RRT∗ extends RRT to find the optimal path, but it needs search every state from initial global scope asymptotically. This method not only inefficient, also contrary single-query RRT. In this paper, a new variant RRT∗-Gb informed presented. goal biasing guide goal. Once an found, and ellipsoidal subset work space can be constructed by refine planning....

Inspired by the 'fine-to-coarse' way-finding strategy that human utilized in process of navigation, paper proposed a fast on-line global path planning algorithm based on regionalized roadmap. First, roadmap(RRM) has multi-layered structure is for representing environments. Then, RRM FTC-A* designed to plan an FTC-route('fine-to-coarse' route) with being fine vicinity yet coarse at distance. This can be applied navigation system mobile robots or vehicles. Finally, simulation and physical...

A simple and effective novel approach for scene understanding is addressed in this paper. Based on bag of visual words (BOVW) model, explicit semantics associated with the object image was embedded into words, then various types integrated, finally SEVI-BOVW (semantics vocabulary integrated words) model constructed. Mean Shift algorithm employed to recognize local region scene. Compared approaches presented literature, proposed here can remove a classification or generative during training...

Purpose The purpose of this paper is to propose a receding horizon control approach for the problem locating unknown wireless sensor networks by using mobile robot. Design/methodology/approach A framework used and consists two levels: one decision level, while other level. In spatiotemporal probability occupancy grid method give possible positions all nodes in networks, where posterior distributions are estimated capturing transient signals. virtual robot designed move along edge obstacles...

Recently, broad applications can be found in optical remote sensing images (ORSI), such as urban planning, military mapping, field survey, and so on. Target detection is one of its important applications. In the past few years, with wings deep learning, target algorithm based on CNN has harvested a breakthrough. However, due to different directions sizes ORSI, it will lead poor performance if for ordinary directly applied. Therefore, how improve object model ORSI thorny. Aiming at solving...

This paper discusses the problem of local path planning in a static-obstacle environment by designing PSO-based receding horizon control approach. In order to avoid obstacles, virtual robot is first designed and moves along boundary obstacles. Then, framework control, cost function proposed where target position are integrated, which implies that mobile robots controlled keep security distance velocity consensus with robots, move toward position. Next, constraints processed particle swarm...

This paper is concerned with the problem of finite-time consensus under directed communication topologies for multi-agent systems double-integrator dynamics. First, a relationship between weights in-edges and ones out-edges each agent established used in analysis. Second, based on our previously proposed controller, conditions framework are derived. Finally, effectiveness illustrated by numerical examples.

This article proposes an experience-based route map continuous learning method and applies it into robot planning navigation. First of all, the framework for navigation is designed, which incorporates four cyclic processes planning, motion, perception, extraction, enabling to constantly learn information road experience obtain improve environment. Besides, a growing-on-demand self-organizing neural network algorithm also proposed. based on growing gas algorithm, but does not require...

For indoor mobile robots, building an accurate and compact map is indispensable prerequisite for different navigation tasks. OctoMap can achieve 3D maps, suitable robot navigation. But in the raw OctoMap, there are many sparse outliers, which affect In this paper, we use RGB-D SLAM system to estimate pose of robots build dense point cloud maps real time. We propose a outliers removal algorithm based on K-Nearest Neighbor Gaussian distribution remove maps. And then, more from filtered perform...

In this paper, a fixed-time rigidity-based formation maneuvering control system with distributed finite-time velocity estimator is studied. The controller of the realized by using local coordinate frames agents without global information. We design estimator, in which desired vector considered to be available part directly, and it can estimated at for each agent. target graph assumed minimally infinitesimally rigid. A distance-based has been presented, proves that 3-D space, could escape...

The field of mobile robotics has seen significant growth regarding the use indoor laser mapping technology, but most two-dimensional Light Detection and Ranging (2D LiDAR) can only scan a plane fixed height, it is difficult to obtain information objects below so inaccurate environmental navigation mis-collision problems easily occur. Although three-dimensional (3D) LiDAR also gradually applied, less used in because more expensive requires large amount memory computation. Therefore, data...

This paper presents a novel auction-based structure to multi-robot coalition formation problem. The structure, which is called Coalition Structure Generation based on Credit Mechanism (CoSGCrM), contains sub-optimal member selection algorithm with an analysis of its soundness and completeness. A credit mechanism introduced reduce the complexity for leader in making decision as well restrict profit-oriented robot bidding coalitions. Simulations are given compare first-price auction results...

An image matching method based on the bag of visual words model is presented, and applied to self-localization mobile body. The local feature extracted by SURF employed produce for every image. "From coarse fine" double-layers indexing structure established grouped set. approach hierarchical-clustering used generate group. Double-layers reduces searching space matching, memory load computational complexity sharply. Based structure, body localization includes two steps: rough location...

According to analysis of motion error based on virtual target tracking in mobile robot, we proposed a simple and effective method for formation control multiple robots corridor or like environments. Taking the posture deviation design law, can make problem be converted into series strategy each robot. A unified algorithm framework was given, then strategies robot were designed. The simulation experiment have demonstrated feasibility validity method.

This paper describes a distributed multi-robot simulation system, named RobSim. RobSim is naturally organized into set of plug-in modules, such as motion module, sensors communication module and map etc. It can be extended combined easily by plunging-in new modules. capable simulating various types which used for testing robot navigation, planning, collaboration or network control, A demonstration application has been developed to validate the RobSimpsilas effectiveness practicality.

The installation of binocular vision at the end a manipulator reduces its availability in environments with obstacles. To deal problem, this study puts forward target localization method using laser and monocular hand-eye vision. In proposed method, centre spot is obtained by vision, geometric relations among emission point, light-spot optical axis camera are used to calculate distance. Then, D-H employed construct coordinate conversion system, so that location can be calculated. measuring...

Self-organizing neural networks are characterized by topology preservation, dynamic adaptation, clustering, and dimensionality reduction, which prompt their wide application in data mining, knowledge extraction, image processing. However, existing self-organizing fail to automatically generate an output space that contains appropriate number of neurons according the input. To address this problem, paper proposes a growing gas (GNG) algorithm with adaptive network scale, is called...