基于多约束条件的机器人抓取策略学习方法

针对机器人在多类别物体不同任务下的抓取决策问题,提出基于多约束条件的抓取策略学习方法.该方法以抓取对象特征和抓取任务属性为机器人抓取策略约束,通过映射人类抓取习惯规划抓取模式,并采用物体方向包围盒(OBB)建立机器人抓取规则,建立多约束条件的抓取模型.利用深度径向基(DRBF)网络模型结合减聚类算法(SCM)实现抓取策略的学习,两种算法的结合旨在提高学习鲁棒性与精确性.搭建以Refiex 1型灵巧手和AUBO六自由度机械臂组成的实验平台,对多类别物体进行抓取实验.实验结果表明,所提出方法使机器人有效学习到对多物体不同任务的最优抓取策略,具有良好的抓取决策能力.     

一种基于高斯核的RBF神经网络学习算法

RBF神经网络中心等参数确定得是否合理将直接影响到RBF神经网络的学习性能。通过有监督学习的方法来确定RBF神经网络的中心等参数是最一般化的方法。在这种方法中,参数的初始化问题是关键问题。文章在分析RBF神经网络映射性能的基础上,提出了中心等参数初始化的一种方法,并借助于梯度下降法给出了RBF神经网络的学习算法。多种实例表明,所给出的学习算法是有效的。该研究为RBF神经网络的广泛应用提供了一定的技术保障。     

Reformulated radial basis neural networks trained by gradientdescent

This paper presents an axiomatic approach for constructing radial basis function (RBF) neural networks. This approach results in a broad variety of admissible RBF models, including those employing Gaussian RBFs. The form of the RBFs is determined by a generator function. New RBF models can be developed according to the proposed approach by selecting generator functions other than exponential ones, which lead to Gaussian RBFs. This paper also proposes a supervised learning algorithm based on gradient descent for training reformulated RBF neural networks constructed using the proposed approach. A sensitivity analysis of the proposed algorithm relates the properties of RBFs with the convergence of gradient descent learning. Experiments involving a variety of reformulated RBF networks generated by linear and exponential generator functions indicate that gradient descent learning is simple, easily implementable, and produces RBF networks that perform considerably better than conventional RBF models trained by existing algorithms     

一种结构自适应的径向基函数神经网络

提出了一种新的结构自适应的径向基函数(RBF)神经网络模型。在该模型中,自组织映射(SOM)神经网络作为聚类网络,采用无监督学习算法对输入样本进行自组织分类,并将分类中心及其对应的权值向量传递给RBF神经网络,分别作为径向基函数的中心和相应的权值向量;RBF神经网络作为基础网络,采用高斯函数实现输入层到隐层的非线性映射,输出层则采用有监督学习算法训练网络的权值,从而实现输入层到输出层的非线性映射。通过对字母数据集进行仿真,表明该网络具有较好的性能。     

一种先验知识引导的基于二阶段渐进网络的自主抓取策略

针对未知不规则物体在堆叠场景下的抓取任务,提出一种基于二阶段渐进网络（two-stage progressive network,TSPN）的自主抓取方法．首先利用端对端策略获取全局可抓性分布,然后基于采样评估策略确定最优抓取配置．将以上2种策略融合,使得TSPN的结构更加精简,显著减少了需评估样本的数量,能够在保证泛化能力的同时提升抓取效率．为了加快抓取模型学习进程,引入一种先验知识引导的自监督学习策略,并利用220种不规则物体进行抓取学习．在仿真和真实环境下分别进行实验,结果表明该抓取模型适用于多物体、堆叠物体、未知不规则物体、物体位姿随机等多种抓取场景,其抓取准确率和探测速度较其他基准方法有明显提升．整个学习过程历时10天,结果表明使用先验知识引导的学习策略能显著加快学习进程．     

A Theoretical Approach of an Intelligent Robot Gripper to Grasp Polygon Shaped Objects

This paper presents an approach for grasp planning and grasp forces optimization of polygon shaped objects. The proposed approach is an intelligent rule-based method that figures out the minimal number of fingers and minimal values of contact forces. These fingers are required to securely grasp a rigid body in the presence of friction and under the action of some external force. This is accomplished by finding optimal contact points on the object boundary along with minimal number of fingers required for achieving the aforementioned goal. Our system handles every object case independently. It generates a rule base for each object based on adequate values of external forces. The system uses the genetic algorithm as its search mechanism, and a rule evaluation mechanism called bucket brigade for the reinforcement learning of the rules. The process mainly consists of two stages; learning then retrieval. Retrievals act on line utilizing previous knowledge and experience embedded in a rule base. If retrievals fail in some cases, learning is presumed until that case is resolved. The algorithm is very general and can be adapted for interface with any object shape. The resulting rule base varies in size according to the degree of difficulty and dimensionality of the grasping problem.     

基于RBF神经网络的抗噪语音识别

针对目前在噪音环境下语音识别系统性能较差的问题,利用RBF神经网络具有最佳逼近性能、训练速度快等特性,分别采用聚类和全监督训练算法,实现了基于RBF神经网络的抗噪语音识别系统。聚类算法的隐含层训练采用K－均值聚类算法,输出层的学习采用线性最小二乘法;全监督算法中所有参数的调整基于梯度下降法,它是一种有监督学习算法,能够选出性能优良的参数。实验表明,在不同的信噪比下,全监督算法较之聚类算法有更高的识别率。     

基于神经网络的PID自整定控制系统

文章介绍了一种应用神经网络技术建立的PID自整定控制系统,给出了系统结构,详细分析了BP神经网络和RBF神经网络的结构和学习算法。该系统采用3层BP神经网络,其输出为PID控制器的参数;通过变结构的RBF神经网络辨识控制对象,将得到的输出对输入的梯度信息提供给BP神经网络,BP神经网络根据该信息优化PID控制器参数。仿真结果表明,该系统对于参数扰动较大的非线性系统,其收敛速度快、动态响应能力强、稳定性好,且具有较强的鲁棒性和适应性。     

gripper

Grasping of objects has been a challenging task for robots. The complex grasping task can be defined as object contact control and manipulation subtasks. In this paper, object contact control subtask is defined as the ability to follow a trajectory accurately by the fingers of a gripper. The object manipulation subtask is defined in terms of maintaining a predefined applied force by the fingers on the object. A sophisticated controller is necessary since the process of grasping an object without a priori knowledge of the object's size, texture, softness, gripper, and contact dynamics is rather difficult. Moreover, the object has to be secured accurately and considerably fast without damaging it. Since the gripper, contact dynamics, and the object properties are not typically known beforehand, an adaptive critic neural network (NN)-based hybrid position/force control scheme is introduced. The feedforward action generating NN in the adaptive critic NN controller compensates the nonlinear gripper and contact dynamics. The learning of the action generating NN is performed on-line based on a critic NN output signal. The controller ensures that a three-finger gripper tracks a desired trajectory while applying desired forces on the object for manipulation. Novel NN weight tuning updates are derived for the action generating and critic NNs so that Lyapunov-based stability analysis can be shown. Simulation results demonstrate that the proposed scheme successfully allows fingers of a gripper to secure objects without the knowledge of the underlying gripper and contact dynamics of the object compared to conventional schemes.     

一种结构自适应径向基函数神经网络

提出了一种新的结构自适应的径向基函数(RBF)神经网络模型。在该网络中,自组织映射(SOM)神经网络作为聚类网络,采用无监督学习算法对输入样本进行自组织分类,并将分类中心及其对应的权值向量传递给RBF神经网络,作为径向基函数的中心和相应的权值向量;RBF神经网络作为基础网络,采用高斯函数实现输入层到隐层的非线性映射,输出层则采用有监督学习算法训练网络的权值,从而实现输入层到输出层的非线性映射。通过对字母数据集进行仿真,表明该网络具有较好的性能。     

Growing radial basis neural networks: merging supervised andunsupervised learning with network growth techniques

This paper proposes a framework for constructing and training radial basis function (RBF) neural networks. The proposed growing radial basis function (GRBF) network begins with a small number of prototypes, which determine the locations of radial basis functions. In the process of training, the GRBF network gross by splitting one of the prototypes at each growing cycle. Two splitting criteria are proposed to determine which prototype to split in each growing cycle. The proposed hybrid learning scheme provides a framework for incorporating existing algorithms in the training of GRBF networks. These include unsupervised algorithms for clustering and learning vector quantization, as well as learning algorithms for training single-layer linear neural networks. A supervised learning scheme based on the minimization of the localized class-conditional variance is also proposed and tested. GRBF neural networks are evaluated and tested on a variety of data sets with very satisfactory results.     

Synergy-based affordance learning for robotic grasping

In this paper, we present an affordance learning system for robotic grasping. The system involves three important aspects: the affordance memory, synergy-based exploration, and a grasping control strategy using local sensor feedback. The affordance memory is modeled with a modified growing neural gas network that allows affordances to be learned quickly from a small dataset of human grasping and object features. After being trained offline, the affordance memory is used in the system to generate online motor commands for reaching and grasping control of the robot. When grasping new objects, the system can explore various grasp postures efficiently in the low dimensional synergy space because the synergies automatically avoid abnormal postures that are more likely to lead to failed grasps. Experimental results demonstrated that the affordance memory can generalize to grasp new objects and predict the effect of the grasp (i.e., the tactile patterns).     

Input Displacement Neuro-fuzzy Control and Object Recognition by Compliant Multi-fingered Passively Adaptive Robotic Gripper

The requirement for new flexible adaptive grippers is the ability to detect and recognize objects in their environments. It is known that robotic manipulators are highly nonlinear systems, and an accurate mathematical model is difficult to obtain, thus making it difficult make decision strategies using conventional techniques. Here, an adaptive neuro fuzzy inference system (ANFIS) for controlling input displacement and object recognition of a new adaptive compliant gripper is presented. The grasping function of the proposed adaptive multi-fingered gripper relies on the physical contact of the finger with an object. This design of the each finger has embedded sensors as part of its structure. The use of embedded sensors in a robot gripper gives the control system the ability to control input displacement of the gripper and to recognize particular shapes of the grasping objects. Fuzzy based controllers develop a control signal according to grasping object shape which yields on the firing of the rule base. The selection of the proper rule base depending on the situation can be achieved by using an ANFIS strategy, which becomes an integrated method of approach for the control purposes. In the designed ANFIS scheme, neural network techniques are used to select a proper rule base, which is achieved using the back propagation algorithm. The simulation results presented in this paper show the effectiveness of the developed method.     

基于径向基函数神经网络的非线性模型辨识

从径向基函数（RBF）神经网络原理分析出发，提出了一种基于RBF神经网络学习算法，用于对非线性对象模型的拟合与辩识，并将此方法用于实际非线性模型的学习与辩识。结果表明，基于RBF的神经网络可快速完成对样本的学习与拟合，对具有连续特性的线性与非线性模型，具有快速实时的学习速度和优良的学习性能。     

基于深度神经网络的多指灵巧手抓取手势优化

基于深度神经网络模型,提出了一种适用于多指灵巧手的抓取手势优化方法。首先,在仿真环境下构建了一个抓取数据集,并在此基础上训练了一个卷积神经网络,依据目标物体单目视觉信息和多指灵巧手抓取位形来预测抓取质量函数,由此可以将多指灵巧手的抓取规划问题转化为使抓取质量最大化的优化问题,进一步,基于深度神经网络中的反向传播和梯度上升算法实现多指灵巧手抓取手势的迭代与优化。在仿真环境中,比较该网络和仿真平台对同一抓取位形的抓取质量评估结果,再利用所提出的优化方法对随机搜索到的初始手势进行优化,比较优化前后手势的力封闭指标。最后,在实际机器人平台上验证本文方法的优化效果,结果表明,本文方法对未知物体的抓取成功率在80%以上,对于失败的抓取,优化后成功的比例达到90%。     

A People-Counting System Using a Hybrid RBF Neural Network

A people-counting system using hybrid RBF neural network is described. The proposed system is effective and flexible for the purpose of performing on-line people counting. Compared with other conventional approach, this system introduces a novel method for feature extraction. In this Letter, a new type of hybrid RBF network is developed to enhance the classification performance. The hybrid RBF based people-counting system is thoroughly compared with other approaches. Extensive and promising results were obtained and the analysis indicates that the proposed hybrid RBF based system provides excellent people-counting results in an open passage. A supervised clustering method is proposed for initialising the hybrid RBF network. In order to substantiate the introduction of the hybrid RBF and the proposed supervised clustering algorithm, test results on a vowel recognition benchmark dataset are also included in the Letter.     

An adiabatic neural network for RBF approximation

Numerous studies have addressed nonlinear functional approximation by multilayer perceptrons (MLPs) and RBF networks as a special case of the more general mapping problem. The performance of both these supervised network models intimately depends on the efficiency of their learning process. This paper presents an unsupervised recurrent neural network, based on the recurrent Mean Field Theory (MFT) network model, that finds a least-squares approximation to an arbitrary L₂ function, given a set of Gaussian radially symmetric basis functions (RBFs). Essential is the reformulation of RBF approximation as a problem of constrained optimisation. A new concept of adiabatic network organisation is introduced. Together with an adaptive mechanism of temperature control this allows the network to build a hierarchical multiresolution approximation with preservation of the global optimisation characteristics. A revised problem mapping results in a position invariant local interconnectivity pattern, which makes the network attractive for electronic implementation. The dynamics and performance of the network are illustrated by numerical simulation.     

Computer aided analysis and derivation for artificial neuralsystems

The theoretical analysis and derivation of artificial neural systems, which consists essentially of manipulating symbolic mathematical objects according to certain mathematical and biological knowledge, can be done more efficiently with computer assistance by using and extending methods and systems of symbolic computation. After presenting the mathematical characteristics of neural systems and a brief review on Lyapunov stability theory, the authors present some features and capabilities of existing systems and the extension for manipulating objects occurring in the analysis of neural systems. Some strategies and a toolkit developed in MACSYMA for computer-aided analysis and derivation are described. A concrete example is given to demonstrate the derivation of a hybrid neural system, i.e. a system which in its learning rule combines elements of supervised and unsupervised learning. Future work and research directions are indicated     

Levenberg-Marquardt算法在T-S型模糊RBF神经网络训练中的应用

为了提高T—S型模糊RBF神经网络的训练效率,把Levenberg—Marquardt算法引入到T—S型模糊RBF神经网络的训练过程中,提高了网络训练的收敛速度,减小了训练过程陷入局部极小点的概率,然后基于这种算法推导出T—S型模糊RBF神经网络的快速训练算法,即混合学习算法。最后通过实验验证了这种算法的有效性和实用性。