In this article, we develop a navigation strategy based on deep reinforcement learning (DRL) for mobile robots. Because of the large difference between simulation and reality, most trained DRL models cannot be directly migrated into real Moreover, how to explore in sparsely rewarded environment is also long-standing problem DRL. This article proposes an end-to-end planner that translates sparse laser ranging results movement actions. Using highly abstract data as input, agents by can...

The objective of visual servoing aims to control an object's motion with feedbacks and becomes popular recently. Problems complex modeling instability always exist in methods. Moreover, there are few research works on selection the gain image-based (IBVS) This paper proposes IBVS method Q-Learning, where learning rate is adjusted by a fuzzy system. Meanwhile, synthetic preprocess introduced perform feature extraction. extraction actually combination color-based recognition algorithm improved...

Image-based visual servoing (IBVS) can reach a desired position for relatively stationary target using continuous feedback. Proper feature extraction and appropriate control laws are essential to performance IBVS. IBVS be interrupted or interfered abruptly if no features extracted when the observed object is occluded. To address problem of missing points in current images during navigation task, homography method that uses priori information proposed predict all ensure execution The mixture...

In a practical continuous-variable quantum key distribution (CVQKD) system, the optical amplifier can be used to improve performance of system by compensating for imperfections detectors. However, eavesdropper Eve actively utilize reverse external magnetic field deteriorate amplifier, which may affect security system. this paper, we investigate CVQKD with an under effect field. Based on investigation parameter estimation above influence, find that evaluation secret rate overestimated, will...

To address the performance bottleneck for image-based visual servoing (IBVS), it is necessary to have appropriate control laws, increased accuracy image feature detection, and minimal approximation errors. This article proposes a fuzzy adaptive method decoupled IBVS that allows efficient of wheeled mobile robot (WMR). under-actuated dynamics WMR, controller used translation rotation are by using two independent gains, instead single gain traditional IBVS. reduce effect noise, this develops...

Some researchers have introduced transfer learning mechanisms to multiagent reinforcement (MARL). However, the existing works devoted cross-task for systems were designed just homogeneous agents or similar domains. This work proposes an all-purpose cross-transfer method, called lateral (MALT), assisting MARL with alleviating training burden. We discuss several challenges in developing method and provide a feasible way of reusing knowledge MARL. In developed we take features as object rather...

Large-scale multiagent reinforcement learning requires huge computation and space costs, the too-long execution process makes it hard to train policies for agents. This work proposes a concept of fuzzy agent, which is new paradigm training homogeneous Aiming at lightweight affordable mechanism large-scale systems, we break one-to-one correspondence between agent policy, designing abstract agents as substitute interact with environment. The Markov decision models these are conducted by logic,...

Robotic soccer games, which have become popular, require timely and precise decisionmaking in a dynamic environment.To address the problems of complexity critical situation, policy improvement robotic games must occur.This paper proposes an adaptive method that uses reinforcement learning (RL), decision-making system for game is composed two subsystems.The first subsystem architecture proposed criticizes second implements policy.Inspired by support vector machine (SVM), situation...

To tackle the problem on trajectory planning or design of control law, this paper introduces a visual servoing system for manipulator with redundant joints that approaching target is determined spontaneously by law. The proposed method resolves joint solution and obstacle avoidance. work comprises two procedures, feature extraction position-based (PBVS) collision avoidance within working envelope. In PBVS control, pose must be reconstructed respect to robot results in Cartesian...

Techniques for transferring human behaviors to robots through learning by imitation/demonstration have been the subject of much study. However, direct transfer motion trajectories humanoid does not result in dynamically stable robot movements because differences and kinematics dynamics. An imitating algorithm called posture-based imitation with balance (Post-BL) is proposed this paper. This Post-BL consists three parts: a key posture identification method used capture postures as knots...

The challenges of selecting appropriate image features, optimizing complex nonlinear computations, and minimizing the approximation errors always exist in visual servoing. A fuzzy neural network controller is developed for a six-degrees-of-freedom robot manipulator to perform servoing proposed tackle these problems. To increase accuracy preprocesses, synthetic process performs feature extraction controller. method combines support vector machine contour recognition algorithm color-based...

Image-based visual servoing (IBVS) allows precise control of positioning and motion for relatively stationary targets using feedback. For IBVS, a mixture parameter β better approximation the image Jacobian matrix, which has significant effect on performance IBVS. However, setting depends camera's realtime posture; there is no clear way to define change rules most IBVS applications. Using simple model-free reinforcement learning, Q-learning, this article proposes method adaptively adjust...

With the explosive growth of Internet Things (IoT) devices and various emerging network technologies, IoT-enabled smart cities are further refined into health cities. For example, IoT can automatically recognize emotional states through collected facial expressions, which serve mental assessment, human–computer interaction, etc. On other hand, existing expression recognition algorithms emphasize application deep neural networks (DNNs), it is difficult for resource-constrained to provide...

Image-based visual servoing (IBVS) achieves precise positioning and motion control for a relatively stationary target by feedback, but problems persist with convergence stability. Appropriate gains the IBVS are critical to stability, this gain is heuristically constant most applications. This paper proposes an integrated method that allows adaptive adjustment of reinforcement learning (RL) control. The proposed learns policy determine value on fly. To ensure rapid RL, truncating Q-learning...

Over the last couple of years, object detection and tracking reserachers have been developing many new techniques, which has used widely by others. In this article, we present an extensive overview methods. At same time, also introduces some related theoretical knowledge (e.g., feature classification). The reason why in summarized together, is because can be said to foundation tracking, they all need choose right features training effective classification. Due application fields emphasis may...

In the reinforcement learning (RL) system, one important issue is tradeoff problem between exploration and exploitation. this paper, we studied dilemma proposed a new approach to solving by multiple-attribute decision making (MADM). The applicability of method extended transfer learning. decomposes task into several subtasks uses policies trained RL. visual MADM (V-MADM) based on state-action values each subtask select action with maximal one. Meanwhile, paper proposes using decay function...