基于最先策略增强学习的ART2神经网络

提出一种基于最先策略增强学习的ART2神经网络FPRL-ART2(Foremost-Policy Reinforcement Learning based ART2 neural network),并介绍其学习算法.为了达到在线学习的目的,在FPRL-ART2中,从状态到行为值之间的映射中,选择第一个得到奖励的行为,而不是选择诸如1-step Q-Learning中具有最优行为值的行为.ART2神经网络用于存储分类模式,其权重通过增强学习增强或减弱,达到学习的目的.并将FPRL-ART2运用到移动机器人避碰撞问题的研究中.仿真实验表明,引入FPRL-ART2后减少移动机器人与障碍物发生碰撞的次数,具有良好的避碰效果.     

一种基于强化学习的ART2神经网络——RL-ART2

提出一种基于强化学习的ART2神经网络(RL-ART2),使其利用强化学习的特性通过与环境交互而无需训练样本即可进行在线学习,同时给出该神经网络的学习算法.当ART2神经网络运行时,通过内部竞争学习得到输出的分类模式,随后通过与环境交互得到神经网络分类模式的运行效果并对其进行评价.通过这种不断与环境的交互学习,当经过在线学习足够的时间和次数后,ART2神经网络即具有相当的识别率.移动机器人路径规划仿真实验表明,使用RL-ART2后与未使用前相比大大减少了机器人与障碍物的碰撞次数,实践证明该方法的合理性和有效性.     

基于速度变化空间的移动机器人动态避碰规划

研究了移动机器人对运动障碍物的动态避碰．针对以往速度障碍法在动态避碰应用中存在的问题,制 订了相应的改进方法．综合考虑障碍物速度的动态变化和碰撞时间、碰撞距离,在速度变化空间中,基于避碰行为 动力学原理,设计了新的优化评价函数,采用双障碍物检测窗口进行动态避碰规划．仿真实验表明,该方法有效地 克服了避碰规划的保守性,提高了机器人运动的安全性,并能实现对运动目标的及时追踪．     

基于双层模糊逻辑的多机器人路径规划与避碰

针对无通信情况下的多机器人系统在未知动态环境下的路径规划问题,设计了基于双层模糊逻辑的多机器人路径规划与动态避碰系统。方向模糊控制器充分考虑了障碍物的距离信息和目标的角度信息,转化为机器人与障碍物的碰撞可能性,从而输出转向角度实现机器人的动态避障;速度模糊控制器将障碍物的距离信息作为输入,将速度因子作为输出,提高了多机器人路径规划与动态避碰系统的效率和鲁棒性。在Pioneer3-DX机器人实体上验证了该系统的可行性。     

CBR船舶避碰决策系统聚类检索的设计

通过在CBR船舶避碰决策系统检索过程中引入ART1神经网络聚类模型,设计CBR系统"子库索引 ART1聚类"的二级检索机制,可以控制检索获得的相似案例集案例数量,提高系统的运行效率.     

基于直觉模糊ART神经网络的群事件检测方法

描述了态势评估系统中的目标编群问题、目标群处理流程和群事件的检测。结合直觉模糊贴近度理论,构造了直觉模糊ART神经网络。设计了网络的运行机制和网络权值向量的学习机制。给出了一个具体实例,检验了直觉模糊ART神经网络的目标编群效果,为群事件检测提供了一条有效途径。     

一种基于效益的多机器人避碰协调策略

多移动机器人系统在完成同时定位和地图构建SLAM任务时,机器人之间常常存在相互碰撞的问题,而这种碰撞的避免又不同于一般的避障,因为避障问题中的障碍物一般是不动的。为了解决机器人之间的避碰问题,提出了一种基于效益的多机器人避碰协调策略。该策略以提高多机器人系统探索效率为主,确定机器人通过交叉路口的顺序。同时考虑了动态协调避碰的情况,给出了确定机器人通过交叉路口顺序的算法。通过机器人在交叉路口实现避碰协调算法的仿真示例,对该方法的避碰协调过程进行了说明,并对仿真结果进行了分析,同时对仿真中机器人和目标位置的空间关系给出了合理的假设。     

基于遗传策略和神经网络的非监督分类方法

文章提出了一种新的基于遗传策略和模糊ART(adaptive resonance theory)神经网络的非监督分类方法.首先,利用原有的训练样本对模糊ART神经网络进行非监督训练,然后,采用遗传策略为模糊ART神经网络增加各类族边界邻域内的训练样本点,再对模糊ART神经网络进行有监督训练.这种方法解决了训练样本在较少条件下的ART系列神经网络的学习与分类问题,提高了ART系列神经网络的分类性能,并扩展了其应用范围.     

ART神经网络的发展与应用

作为解决神经网络学习中“稳定性/可塑性两难问题“的一种尝试,ART神经网络一直备受关注.从最初的仅仅用于处理二值输入的非监督学习网络ART1,到具有有监督学习能力的ARTMAP网络,具有一定模糊逻辑运算能力的Fuzzy ART网络,再到现在对于ART网络中的各种尝试,ART神经网络不断发展、改进,以便适应不同的应用场合.本文着重介绍了ART网络的基本体系结构与发展历程,对于其应用领域加以概述.     

基于ART2神经网络的车牌字符识别

提出了ART2神经网络在车牌自动识别系统中的应用.ART2神经网络可以在非平稳的、有干扰的环境中进行无教师无监督的自学习,学习过程是自组织的实时学习,能迅速识别已学习过的样本,并能迅速适应未学习过的新对象.Zernike矩具有旋转不变性、位移不变性、比例不变性.该方法结合了Zernike矩和ART2神经网络的优点,在实验中取得了很好的效果,解决了车牌自动识别系统中字符识别的难题.     

Navigation system including associative memory

A new adaptive linear robot control system for a robot work cell that can visually track and intercept stationary and moving objects undergoing arbitrary motion anywhere along its predicted trajectory within the robot's workspace is presented in this paper. The proposed system was designed by integrating a stationary monocular CCD camera with off-the-shelf frame grabber and an industrial robot operation into a single application on the MATLAB platform. A combination of the model based object recognition technique and a learning vector quantization network is used for classifying stationary objects without overlapping. The optical flow technique and the MADALINE network are used for determining the target trajectory and generating the predicted robot trajectory based on visual servoing, respectively. The necessity of determining a model of the robot, camera, all the stationary and moving objects, and environment is eliminated. The location and image features of these objects need not be preprogrammed, marked and known before, and any change in a task is possible without changing the robot program. After the learning process on the robot, it is shown that the KUKA robot is capable of tracking and intercepting both stationary and moving objects at an optimal rendezvous point on the conveyor accurately in real-time.     

基于ART2的Q学习算法研究

为了解决Q学习应用于连续状态空间的智能系统所面临的"维数灾难"问题,提出一种基于ART2的Q学习算法.通过引入ART2神经网络,让Q学习Agent针对任务学习一个适当的增量式的状态空间模式聚类,使Agent无需任何先验知识,即可在未知环境中进行行为决策和状态空间模式聚类两层在线学习,通过与环境交互来不断改进控制策略,从而提高学习精度.仿真实验表明,使用ARTQL算法的移动机器人能通过与环境交互学习来不断提高导航性能.     

基于情感与环境认知的移动机器人自主导航控制

将基于情感和认知的学习与决策模型引入到基于行为的移动机器人控制体系中, 设计了一种新的自主导航控制系统. 将动力学系统方法用于基本行为设计, 并利用ART2神经网络实现对连续的环境感知状态的分类, 将分类结果作为学习与决策算法中的环境认知状态. 通过在线情感和环境认知学习, 形成合理的行为协调机制. 仿真表明, 情感和环境认知能明显地改善学习和决策过程效率, 提高基于行为的移动机器人在未知环境中的自主导航能力     

Learning sensor-based navigation of a real mobile robot in unknownworlds

In this paper, we address the problem of navigating an autonomous mobile robot in an unknown indoor environment. The parti-game multiresolution learning approach is applied for simultaneous and cooperative construction of a world model, and learning to navigate through an obstacle-free path from a starting position to a known goal region. The paper introduces a new approach, based on the application of the fuzzy ART neural architecture, for on-line map building from actual sensor data. This method is then integrated, as a complement, on the parti-game world model, allowing the system to make a more efficient use of collected sensor information. Then, a predictive on-line trajectory filtering method, is introduced in the learning approach. Instead of having a mechanical device moving to search the world, the idea is to have the system analyzing trajectories in a predictive mode, by taking advantage of the improved world model. The real robot will only move to try trajectories that have been predicted to be successful, allowing lower exploration costs. This results in an overall improved new method for goal-oriented navigation. It is assumed that the robot knows its own current world location-a simple dead-reckoning method is used for localization in our experiments. It is also assumed that the robot is able to perform sensor-based obstacle detection (not avoidance) and straight-line motions. Results of experiments with a real Nomad 200 mobile robot are presented, demonstrating the effectiveness of the discussed methods.     

目标围捕任务中搜索与预包围阶段的群机器人行为学习

为了进行群机器人协同作业,提出目标搜索中导航类集体行为学习策略.在使用具有闭环调节功能的动态任务分工方法进行任务分配、自组织地生成多个子群后,在子群中引入基于社会学习微粒群算法的机器人行为学习策略.在子群框架内,机器人各自独立地以感知的共同意向目标信号强度为标准对所有成员排序,将感知优于自己的机器人作为行为示范者.然后在搜索空间各维度上分别随机选择一个行为示范者,学习其在相应维度上的位置坐标,经构造得到搜索空间中自己的学习行为向量,由此决策自身的运动行为.仿真结果表明,在不需要学习全局社会经验的前提下,机器人能针对所属子群的共同意向目标进行协同作业,提高搜索效率.     

Conditions for target tracking with range-only information

This paper addresses the problem of guiding a mobile robot towards a target using only range sensors. The bearing information is not available. The target can be stationary or moving. It can be the source of some gas leakage or nuclear radiation or it can be some landmark or beacon or any manoeuvring vehicle. The mobile robot can be a ground vehicle or an aerial vehicle flying at a fixed altitude. In literature, many different strategies are proposed which use the range only measurement but they involve estimation of different parameters or have switching control strategy which make them difficult to implement. We propose two sets of conditions, one for stationary target and another for both stationary and moving target. Any control strategy, that will satisfy these conditions, can bring the robot arbitrarily close to the target. There are no restrictions on the initial conditions. Estimation of any parameter is not required. Some candidate controllers are presented that included continuous controllers and switching controllers. Simulations are carried out with these controllers to validate our result with and without measurement noise. Experimental results with ground mobile robot are presented.     

Design and integration of a spatio-temporal memory with emotional influences to categorize and recall the experiences of an autonomous mobile robot

Autonomous robots cohabiting with humans will have to achieve recurring tasks while adapting to the changing conditions of the world. A spatio-temporal memory categorizes the experiences of a robot to improve its ability to adapt to its environment. In this paper, we present a spatio-temporal (ST) memory model consisting of a cascade of two adaptive resonance theory (ART) networks: one to categorize spatial events and the other to extract temporal episodes from the robot’s experiences. Artificial emotions are used to dynamically modulate learning and recall of the ART networks based on how the robot is able to carry its task, using a simple model of artificial emotions. Once an episode is recalled, future events can be predicted and used to influence the intentions of the robot. Evaluation of our ST model is done using an autonomous robotic platform that has to deliver objects to people within an office area. Results demonstrate that our model can memorize and recall the experiences of a robot, and that emotional episodes are recalled more often, allowing the robot to use its memory of past experiences early on when repeating a task.     

Real-Time Dynamic Visual Tracking Using PSD Sensors and Extended Trapezoidal Motion Planning

A real-time visual servo tracking system for an industrial robot has been implemented using PSD (Position Sensitive Detector) cameras, neural networks, and an extended trapezoidal motion planning method. PSD and directly transduces the light's projected position on its sensor plane into an analog current and lends itself to fast real-time tracking. A neural network, after proper training, transforms the PSD sensor reading into a 3D position of the target, which is then input to an extended trapezoidal motion planning algorithm. This algorithm implements a continuous motion update strategy in response to an ever-changing sensor information from the moving target, while greatly reducing the tracking delay. This planning method is found to be very useful for sensor-based control such as moving target tracking or weld-seam tracking in which the robot needs to change its motion in real time in response to incoming sensor information. Further, for real-time usage of the neural net, a new architecture called LANN (Locally Activated Neural Network) has been developed based on the concept of CMAC input partitioning and local learning. Experimental evidence shows that an industrial robot can smoothly track a moving target of unknown motion with speeds of up to 1 m/s and with oscillation frequency up to 5 Hz.     

用于模式识别的ART-2神经网络算法的改进

针对模式识别中模式有序输出的要求,对ＡＲＴ－２神经网络的算法进行了改进和调整,提出了ＡＲＴ－２神经网络的改进算法,通过对改进算法与原算法的识别试验结果进行比较,表明该改进算法对模式的有序输出是可行的和有效的。     

Learning Inverse Kinematics: Reduced Sampling Through Decomposition Into Virtual Robots

We propose a technique to speedup the learning of the inverse kinematics of a robot manipulator by decomposing it into two or more virtual robot arms. Unlike previous decomposition approaches, this one does not place any requirement on the robot architecture, and thus, it is completely general. Parametrized self-organizing maps are particularly adequate for this type of learning, and permit comparing results directly obtained and through the decomposition. Experimentation shows that time reductions of up to two orders of magnitude are easily attained.