Rapid Concept Learning for Mobile Robots

Concept learning in robotics is an extremely challenging problem: sensory data is often high-dimensional, and noisy due to specularities and other irregularities. In this paper, we investigate two general strategies to speed up learning, based on spatial decomposition of the sensory representation, and simultaneous learning of multiple classes using a shared structure. We study two concept learning scenarios: a hallway navigation problem, where the robot has to induce features such as opening or wall. The second task is recycling, where the robot has to learn to recognize objects, such as a trash can. We use a common underlying function approximator in both studies in the form of a feedforward neural network, with several hundred input units and multiple output units. Despite the high degree of freedom afforded by such an approximator, we show the two strategies provide sufficient bias to achieve rapid learning. We provide detailed experimental studies on an actual mobile robot called PAVLOV to illustrate the effectiveness of this approach.     

再励学习及其在移动机器人行为规划中的应用

再励学习（Reinforcement Learning,RL）是一种成功地结合动态编程和控制问题的机器智能方法,它将动态编程和有监督学习方法结合到机器学习系统中,通常用于解决预测和控制两类问题。提出了以矢量形式表示的评估函数,为了实现多维再励学习,用一专门的神经网络（Q网络）实现评判网络,研究其在移动机器人行为规划中的应用。     

Connectionist Learning in Behaviour-Based Mobile Robots: A Survey

This paper is a survey of some recentconnectionist approaches to the design and developmentof behaviour-based mobile robots. The research isanalysed in terms of principal connectionist learningmethods and neurological modeling trends. Possibleadvantages over conventionally programmed methods areconsidered and the connectionist achievements to dateare assessed. A realistic view is taken of theprospects for medium term progress and someobservations are made concerning the direction thismight profitably take.     

基于强化学习和模糊逻辑的移动机器人导航

自主导航是移动机器人的一项关键技术。该文采用强化学习结合模糊逻辑的方法实现了未知环境下自主式移动机机器人的导航控制。文中首先介绍了强化学习原理,然后设计了一种未知环境下机器人导航框架。该框架由避碰模块、寻找目标模块和行为选择模块组成。针对该框架,提出了一种基于强化学习和模糊逻辑的学习、规划算法：在对避碰和寻找目标行为进行独立学习后,利用超声波传感器得到的环境信息进行行为选择,使机器人在成功避碰的同时到达目标点。最后通过大量的仿真实验,证明了算法的有效性。     

A Multiple Models Approach for Adaptation and Learning in Mobile Robots Control

The paper proposes a multiple models based control methodology for the solution of the tracking problem for mobile robots. The proposed method utilizes multiple models of the robot for its identification in an adaptive and learning control framework. Radial Basis Function Networks (RBFNs) are considered for the multiple models in order to exploit the non-linear approximation capabilities of the nets for modeling the kinematic behaviour of the vehicle and for reducing unmodelled tracking errors contributions. The training of the nets and the control performance analysis have been done in a real experimental setup. The experimental results are satisfactory in terms of tracking errors and computational efforts and show the improvement in the tracking performance when the proposed methodology is used for tracking tasks in dynamical uncertain environments.     

基于深度学习的移动机器人自主导航实验平台

随着移动机器人的发展,其应用场景越来越复杂,对自主导航这一关键技术提出了更高要求。本文搭建了移动机器人实验平台,设计了基于深度学习的自主导航方法,将RGB图像作为卷积神经网络模型的输入,即可直接输出导航控制信号,不仅降低硬件成本,而且避免复杂的特征工程和规划策略。实验结果表明该平台具有良好的自主导航性能,对移动机器人适应未知复杂环境作业有着重要参考价值。同时,能够为机器人工程专业实践教学提供实验平台,通过开展相关应用拓展,促进学生创新研究能力的培养。     

基于神经网络的移动机器人路径规划

针对移动机器人未知环境下的安全路径规划,本文采用了一种局部连接Hopfield神经网络（ANN）规划器。对任意形状环境,ANN中兼顾处理了“过近”和“过远”来形成安全 路径,而无需学习过程。为在单处理器上进行有效的在线路径规划,提出用基于距离变换的串行模拟,加速数值势场的传播。仿真表明,该方法具有较高的实时性和环境适应性。     

Synthesis of Neural Networks and PID Control for Performance Improvement of Industrial Robots

In this article, an approach for improving the performance of industrialrobots using multilayer feedforward neural networks is presented. Thecontroller based on this approach consists of two main components: a PIDcontrol and a neural network. The function of the neural network is tocomplement the PID control for the specific purpose of improving theperformance of the system over time. Analytical and experimental resultsconcerning this synthesis of neural networks and PID control are presented.The analytical results assert that the performance of PID-controlledindustrial robots can be improved through proper utilization of the learningand generalization ability of neural networks. The experimental results,obtained through actual implementation using a commercial industrial robot,demonstrate the effectiveness of such control synthesis for practicalapplications. The results of this work suggest that neural networks could beadded to existing PID-controlled industrial robots for performanceimprovement.     

基于神经网络的机器人学习与控制：回顾与展望

机器人因其高效的感知、决策和执行能力,在人工智能、信息技术和智能制造等领域中具有巨大的应用价值。目前,机器人学习与控制已成为机器人研究领域的重要前沿技术之一。各种基于神经网络的智能算法被设计,从而为机器人系统提供同步学习与控制的规划框架。首先从神经动力学(ND)算法、前馈神经网络(FNNs)、递归神经网络(RNNs)和强化学习(RL)四个方面介绍了基于神经网络的机器人学习与控制的研究现状,回顾了近30年来面向机器人学习与控制的智能算法和相关应用技术。最后展望了该领域存在的问题和发展趋势,以期促进机器人学习与控制理论的推广及应用场景的拓展。     

Learning from History for Behavior-Based Mobile Robots in Non-Stationary Conditions

Learning in the mobile robot domain is a very challenging task, especially in nonstationary conditions. The behavior-based approach has proven to be useful in making mobile robots work in real-world situations. Since the behaviors are responsible for managing the interactions between the robots and its environment, observing their use can be exploited to model these interactions. In our approach, the robot is initially given a set of behavior-producing modules to choose from, and the algorithm provides a memory-based approach to dynamically adapt the selection of these behaviors according to the history of their use. The approach is validated using a vision- and sonar-based Pioneer I robot in nonstationary conditions, in the context of a multirobot foraging task. Results show the effectiveness of the approach in taking advantage of any regularities experienced in the world, leading to fast and adaptable specialization for the learning robot.     

Learning from History for Behavior-Based Mobile Robots in Non-Stationary Conditions

Learning in the mobile robot domain is a very challenging task, especially in non-stationary conditions. The behavior-based approach has proven to be useful in making mobile robots work in real-world situations. Since the behaviors are responsible for managing the interactions between the robots and its environment, observing their use can be exploited to model these interactions. In our approach, the robot is initially given a set of behavior-producing modules to choose from, and the algorithm provides a memory-based approach to dynamically adapt the selection of these behaviors according to the history of their use. The approach is validated using a vision- and sonar-based Pioneer I robot in non-stationary conditions, in the context of a multi-robot foraging task. Results show the effectiveness of the approach in taking advantage of any regularities experienced in the world, leading to fas t and adaptable specialization for the learning robot.     

Training a Vision Guided Mobile Robot

This paper presents the design, implementation and evaluation of a trainable vision guided mobile robot. The robot, CORGI, has a CCD camera as its only sensor which it is trained to use for a variety of tasks. The techniques used for training and the choice of natural light vision as the primary sensor makes the methodology immediately applicable to tasks such as trash collection or fruit picking. For example, the robot is readily trained to perform a ball finding task which involves avoiding obstacles and aligning with tennis balls. The robot is able to move at speeds up to 0.8 ms^-1 while performing this task, and has never had a collision in the trained environment. It can process video and update the actuators at 11 Hz using a single $20 microprocessor to perform all computation. Further results are shown to evaluate the system for generalization across unseen domains, fault tolerance and dynamic environments.     

基于神经动力学的非完整移动机器人跟踪控制

主要研究了非完整移动机器人轨迹跟踪问题。基于后退控制和神经动力学生物激励模型,采用自适应参数调节的方法提出了一种新的跟踪控制器。该控制器能够生成平稳合理的速度,解决了以往大部分跟踪控制器所产生的速度突变问题,并且具有很好的鲁棒性。运用李雅普诺夫稳定性理论证明控制系统的稳定性。对连续、离散轨迹的仿真及与传统后退方法的比较分析验证了该方法的有效性。     

Training a Vision Guided Mobile Robot

This paper presents the design, implementation and evaluation of a trainable vision guided mobile robot. The robot, CORGI, has a CCD camera as its only sensor which it is trained to use for a variety of tasks. The techniques used for train ing and the choice of natural light vision as the primary sensor makes the methodology immediately applicable to tasks such as trash collection or fruit picking. For example, the robot is readily trained to perform a ball finding task which involves avoiding obstacles and aligning with tennis balls. The robot is able to move at speeds up to 0.8 ms^-1 while performing this task, and has never had a collision in the trained environment. It can process video and update the actuators at 11 Hz using a single $20 microprocessor to perform all computation. Further results are shown to evaluate the system for generalization across unseen domains, fault tolerance and dynamic environments.     

Global Level Path Planning for Mobile Robots in Dynamic Environments

This paper presents a self-adapting approach to global level path planning in dynamic environments. The aim of this work is to minimize risk and delays in possible applications of mobile robots (e.g., in industrial processes). We introduce a hybrid system that uses case-based reasoning as well as grid-based maps for decision-making. Maps are used to suggest several alternative paths between specific start and goal point. The casebase stores these solutions and remembers their characteristics. Environment representation and casebase design are discussed. To solve the problem of exploration vs. exploitation, a decision-making strategy is proposed that is based on the irreversibility of decisions. Forgetting strategies are discussed and evaluated in the context of case-based maintenance. The adaptability of the system is evaluated in a domain based on real sensor data with simulated occupancy probabilities. Forgetting strategies and decision-making strategies are evaluated in simulated environments. Experiments show that a robot is able to adapt in dynamic environments and can learn to use paths that are less risky to follow.     

基于强化学习规则的两轮机器人自平衡控制

两轮机器人是一个典型的不稳定,非线性,强耦合的自平衡系统,在两轮机器人系统模型未知和没有先验经验的条件下,将强化学习算法和模糊神经网络有效结合,保证了函数逼近的快速性和收敛性,成功地实现两轮机器人的自学习平衡控制,并解决了两轮机器人连续状态空间和动作空间的强化学习问题;仿真和实验表明:该方法不仅在很短的时间内成功地完成对两轮机器人的平衡控制,而且在两轮机器人参数变化较大时,仍能维持两轮机器人的平衡。     

一类受限机器人的学习控制

本文给出了一类受限机器人的一个学习控制方案，学习控制器的设计是基于受限机器人的奇异模型，在存在未知的有界干扰的情况下，对末端操纵器受线性、无摩擦约束面约束的机器人，本文给出的控制方案实现了机器人运动的完全跟踪，保证了力跟踪误差是有界的，并且界的大小是可调节的。     

Situated Learning of a Behavior-Based Mobile Robot Path Planner

1TaskDecompositionWebuildourpathplannerbaJsedonBrook'ssubsumptionarchitecturel1].InBrook'sapproach,theoveralltaskisdecomposedilltoseveralconcurrelitbehaviors,eachbehaviorhasitsownaPplicabilityconditionsspecifyingwhenitisappropriate,andapriorityorderingispresdefinedtoresolveconflictsamongbehaviors.Inpathplannillg,thetaskoftherobotistoaPproachatargetwhileavoidingobstacles.Wedecomposethetaskintothreebehaviors.TheyaJreAvoid,SteerandAdvance.Fig.1illustratestheoverallstructureofourpathplanner.…     

基于在线迁移学习的重现概念漂移数据流分类

随着大数据时代的到来,数据流分类被应用于诸多领域,如：垃圾邮件过滤、市场预测及天气预报等.重现概念是这些应用领域的重要特点之一.针对重现概念的学习与分类问题中的“负迁移”和概念漂移检测的滞后性,提出了一种基于在线迁移学习的重现概念漂移数据流分类算法——RC-OTL.RC-OTL在检测到概念漂移时存储刚学习的一个基分类器,然后计算最近的样本与存储的各历史分类器之间的领域相似度,以选择最适合对后续样本进行学习的源分类器,从而改善从源领域到目标领域的知识迁移.另外,RC-OTL还在概念漂移检测之前根据分类准确率选择合适的分类器对后续样本分类.初步的理论分析解释了RC-OTL为什么能有效克服“负迁移”,实验结果进一步表明：RC-OTL的确能有效提高分类准确率,并且在遭遇概念漂移后能更快地适应后续样本.     

一种结合机器学习的移动边缘计算的切换预测方法

随着移动通信的发展,减少通信延时成为关键性问题,因此,提出了一种使用机器学习方法的移动边缘计算(MEC)移动性管理。移动性决策基于参考信号接收功率(RSRP)值和不确定性预测器。使用神经网络建立预测器,输出不同相邻单元的RSRP平均值和标准偏差,推导了切换概率的封闭表达式。基于这些可能性,MEC服务器能够提前缓存用户服务,以便将切换期间的中断降至最低。实验结果表明,提出的方法能够满足实际需求。