A method to learn the inverse kinematics of multi-link robots by evolving neuro-controllers

A general method to learn the inverse kinematic of multi-link robots by means of neuro-controllers is presented. We can find analytical solutions for the most used and well-known robots in the literature. However, these solutions are specific to a particular robot configuration and are not generally applicable to other robot morphologies. The proposed method is general in the sense that it is independent of the robot morphology. The method is based on the evolutionary computation paradigm and works obtaining incrementally better neuro-controllers. Furthermore, the proposed method solves some specific issues in robotic neuro-controller learning: it avoids any neural network learning algorithm which relies on the classical supervised input-target learning scheme and hence it lets to obtain neuro-controllers without providing targets. It can converge beyond local optimal solutions, which is one of the main drawbacks of some neural network training algorithms based on gradient descent when applied to highly redundant robot morphologies. Furthermore, using learning algorithms such as the neuro-evolution of augmenting topologies it is also possible to learn the neural network topology which is a common source of empirical testing in neuro-controllers design. Finally, experimental results are provided when applying the method to two multi-link robot learning tasks and a comparison between structural and parametric evolutionary strategies on neuro-controllers is shown.     

基于降阶模型的斜拉索振动的半主动神经网络控制

根据模态降阶理论,获得了斜拉索-阻尼器系统的降阶模型,有效地缩减了系统的自由度.根据ER/MR阻尼器特性和主动控制中LQG控制理论,建立了面向速度剪切的半主动LQG控制方法,并获得了很好的控制效果.本文设计了神经网络的观测器,使用的传感器数目大大减少,根据智能控制理论,设计了神经网络控制器,并提出采用该神经网络作斜拉索半主动控制的控制算法.振动仿真的结果表明,经过离线训练后的神经网络观测器和控制器,有效地抑制了斜拉索的振动.     

Adaptive critic motion control design of autonomous wheeled mobile robot by dual heuristic programming

Autonomous wheeled mobile robot (WMR) needs implementing velocity and path tracking control subject to complex dynamical constraints. Conventionally, this control design is obtained by analysis and synthesis or by domain expert to build control rules. This paper presents an adaptive critic motion control design, which enables WMR to autonomously generate the control ability by learning through trials. The design consists of an adaptive critic velocity control loop and a self-learning posture control loop. The neural networks in the velocity neuro-controller (VNC) are corrected with the dual heuristic programming (DHP) adaptive critic method. Designer simply expresses the control objective by specifying the primary utility function then VNC will attempt to fulfill it through incremental optimization. The posture neuro-controller (PNC) learns by approximating the specialized inverse velocity model of WMR so as to map planned positions to suitable velocity commands. Supervised drive supplies variant velocity commands for PNC and VNC to set up their neural weights. During autonomous drive, while PNC halts learning VNC keeps on correcting its neural weights to optimize the control performance. The proposed design is evaluated on an experimental WMR. The results show that the DHP adaptive critic design is a useful base of autonomous control.     

Parallelization of the Gaussian Elimination Algorithm on Systolic Arrays

We study the parallel implementation of the Gaussian elimination scheme on systolic arrays. We first show that the time (resp. area) complexity of the algorithm isT= 3n− 1 (resp.S= (n²/4) +O(n)), wherenis the size of the linear system. Then we exhibit three algorithms. The two first ones are optimal in time. The first one corresponds to an orthogonally connected array of size 3(n²/8) +O(n). The second network is smaller,S= 3(n²/10) +O(n), but two layers are necessary in order to obtain a regular layout with local communications. The third one is hexagonally connected, has size (n²/3) +O(n), and is almost optimal in time.     

Gossiping by processors prone to omission failures

We consider the gossip problem in a synchronous message-passing system. Participating processors are prone to omission failures, that is, a faulty processor may fail to send or receive a message. The gossip problem in the fault-tolerant setting is defined as follows: every correct processor must learn the initial value of any other processor, unless the other one is faulty; in the latter case either the initial value or the information about the fault must be learned. We develop two efficient algorithms that solve the gossip problem in time O(logn), where n is the number of processors in the system. The first one is an explicit algorithm (i.e., constructed in polynomial time) sending O(nlogn+f²) messages, and the second one reduces the message complexity to O(n+f²), where f is the upper bound on the number of faulty processors.     

Evolving Neuro-Controllers for a Dynamic System Using Structured Genetic Algorithms

This paper describes the application of the Structured Genetic Algorithm (sGA) to design neuro-controllers for an unstable physical system. In particular, the approach uses a single unified genetic process to automatically evolve complete neural nets (both architectures and their weights) for controlling a simulated pole-cart system. Experimental results demonstrate the effectiveness of the sGA-evolved neuro-controllers for the task—to keep the pole upright (within a specified vertical angle) and the cart within the limits of the given track.     

Numerical differentiation of implicitly defined space curves

In this note we develop a simple finite differencing device to calculate approximations of derivativesx′(0),x″(0),x ⁽³⁾(0), … of regular solution curvesx: ? ?s →x(s) ∈ ? ⁿ of nonlinear systems of equationsg(x)=0,g∈C ^k (? ^{n + 1}, ? ⁿ ) without having to compute points on the solution arcx(s). The derivative vectorsx′(0),x″(0),x ⁽³⁾(0),… can be used in the numerical approximation of the solution setg ^?1(0) in two ways. On one hand they can be applied to construct higher order predictors to be used in predictor-corrector branch following procedures. On the other they serve as order determining basis functions in the Reduced Basis Method. The performance of the differencing method is demonstrated by some numerical examples.     

Bounded neuro-control position regulation for a geared DC motor

The purpose of this paper is to present a simple neuro-control law in order to control a geared DC motor. The main advantage of this controller is that it does not require an exact knowledge of the values of the motor parameters. The proposed artificial neural network is characterized by two input synaptic weights, two output synaptic weights and one threshold; these parameters are used to define the performance of the closed loop system. The DC motor parameters, the synaptic weights and the ANN threshold are combined in order to construct an off-line learning condition. Such condition guarantees that the seminorm of the regulation error remains bounded (closed loop performance index) and it is constructed through a Lyapunov-like analysis. The neuro-controller is evaluated through numerical simulations and through small-scale laboratory experiments by implementing the neuro-controller with electronic hardware.     

An efficient Chebyshev-Tau spectral method for Ginzburg-Landau-Schrödinger equations

We propose an efficient time-splitting Chebyshev-Tau spectral method for the Ginzburg-Landau-Schrödinger equation with zero/nonzero far-field boundary conditions. The key technique that we apply is splitting the Ginzburg-Landau-Schrödinger equation in time into two parts, a nonlinear equation and a linear equation. The nonlinear equation is solved exactly; while the linear equation in one dimension is solved with Chebyshev-Tau spectral discretization in space and Crank-Nicolson method in time. The associated discretized system can be solved very efficiently since they can be decoupled into two systems, one for the odd coefficients, the other for the even coefficients. The associated matrices have a quasi-tridiagonal structure which allows a direction solution to be obtained. The computation cost of the method in one dimension is O(Nlog(N)) compared with that of the non-optimized one, which is O(N²). By applying the alternating direction implicit (ADI) technique, we extend this efficient method to solve the Ginzburg-Landau-Schrödinger equation both in two dimensions and in three dimensions, respectively. Numerical accuracy tests of the method in one dimension, two dimensions and three dimensions are presented. Application of the method to study the semi-classical limits of Ginzburg-Landau-Schrödinger equation in one dimension and the two-dimensional quantized vortex dynamics in the Ginzburg-Landau-Schrödinger equation are also presented.     

Implementation of indirect neuro-control for a nonlinear two-robot MIMO system

This paper presents identification and control designs using neural networks for a nonlinear two-robot multiinput multioutput (MIMO) system. The proposed neuro-controller is a combination of linear controllers and a neural-network controller, and is trained by an indirect neuro-control scheme. The proposed neuro-controller is implemented and tested on an IBM PC-based two 2-bar systems holding an object, and is applicable to many d.c.-motor-driven precision nonlinear two-robot MIMO systems. The algorithm and experimental results are described. The experimental results are shown to be superior to those of conventional control.     

Analyzing the understandability of Requirements Engineering languages for CSCW systems: A family of experiments

ContextA collaborative system is a special kind of software whose users can perform collaboration, communication and collaboration tasks. These systems usually have a high number of non-functional requirements, resulting from the users’ need of being aware of other users with whom to collaborate, that is, the workspace awareness.ObjectiveThis paper aims at evaluating two Requirements Engineering languages i^* and CSRML (an extension of i^*) in order to determine which is the most suitable one to specify requirements of collaborative systems, taking into account their special characteristics regarding collaboration and awareness.MethodWe performed a family of experiments comprising an original experiment and two replicas. They were performed by 30, 45 and 9 Computer Science students, respectively, from Spain and Argentina. These subjects filled in two understandability questionnaires once they analyzed the requirements models of two systems: an e-learning collaborative system and a conference review system with some collaborative aspects support. Both models were specified by using the evaluated languages.ResultsThe statistical analysis of the family of experiments showed that the understandability was higher for the models specified with CSRML than for those specified with i^*, especially for collaborative aspects. This result was also confirmed by the meta-analysis conducted.ConclusionsCSRML surpasses i^* when modeling collaborative systems requirements models due to the specific expressiveness introduced to represent collaboration between users and awareness and the new resorts to manage actors and roles.     

A trigonometrically-fitted one-step method with multi-derivative for the numerical solution to the one-dimensional Schrödinger equation

In this paper we present a new multi-derivative or Obrechkoff one-step method for the numerical solution to an one-dimensional Schrödinger equation. By using trigonometrically-fitting method (TFM), we overcome the traditional Obrechkoff one-step method (or called as the non-TFM) for its poor-accuracy in the resonant state. In order to demonstrate the excellent performance for the resonant state, we consider only the simplest TFM, of which the local truncation error (LTE) is of O(h⁷), a little higher than the one of the traditional Numerov method of O(h⁶), and only the first- and second-order derivatives of the potential function are needed. In the new method, in order to solve two unknowns, wave function and its first-order derivative, we use a pair of two symmetrically linear-independent one-step difference equations. By applying it to the well-known Woods-Saxon's potential problem, we find that the TFM can surpass the non-TFM by five orders for the highest resonant state, and surpass Numerov method by eight orders. On the other hand, because of the small error constant, the accuracy improvement to the ground state is also remarkable, and the numerical result obtained by TFM can be four to five orders higher than the one by Numerov method.     

Experiments with free concept generation in Divago

This paper presents a set of experiments we carried out with, Divago, a system that is an attempt to implement our ideas towards a computational model of creativity. It is expected to be able to generate novel concepts out of previous knowledge. Here we show its behaviour with a large dataset constructed independently by other researchers consisting of over 170 nouns (for a project named C³). Each noun is represented with a syntax that is equivalent to the one adopted for Divago. We apply a two step experimentation procedure, which starts by “training” the system with “preferred outcomes” and then allowing it to do free generation, constrained by the pragmatic goal of a given query. We evaluate the results and make a short discussion regarding well-defined criteria of novelty and usefulness. We also present a comparison with a similar experiment done with C³.     

An optimal tracking neuro-controller for nonlinear dynamic systems

Multilayer neural networks are used to design an optimal tracking neuro-controller (OTNC) for discrete-time nonlinear dynamic systems with quadratic cost function. The OTNC is made of two controllers: feedforward neuro-controller (FFNC) and feedback neuro-controller (FBNC). The FFNC controls the steady-state output of the plant, while the FBNC controls the transient-state output of the plant. The FFNC is designed using a novel inverse mapping concept by using a neuro-identifier. A generalized backpropagation-through-time (GBTT) algorithm is developed to minimize the general quadratic cost function for the FBNC training. The proposed methodology is useful as an off-line control method where the plant is first identified and then a controller is designed for it. A case study for a typical plant with nonlinear dynamics shows good performance of the proposed OTNC.     

Motion synthesis with decoupled parameterization

In real-time animation systems, motion interpolation techniques are widely used for their controllability and efficiency. The techniques sample the parameter space using example motions, and interpolate them to compute the blend weights corresponding to the given parameters. A main problem of the techniques is that, as the dimension n of the parameter space increases, the number of required example motions increases exponentially, i.e. O(c ⁿ ). To resolve the problem, this paper proposes to use two decoupled parameter spaces for controlling the upper body and the lower body separately. At each frame time, a parameterized motion space produces a source frame, and the target frame is synthesized by splicing the upper body of one source frame with the lower body of the other. In order to have the two source frames correlated with each other, a time-warping scheme has been developed. Furthermore, in order to handle the dynamic properties of the parameter samples of the upper body, we have developed an approximation technique for quickly determining the sample positions in its parameter space. This decoupled parameterization method alleviates the complexity problem, e.g. from O(c ⁶) to O(c ³), while providing the users with the capability of convenient control over the character.     

Robust parametric CMAC with self-generating design for uncertain nonlinear systems

In this paper, a robust parametric cerebellar model articulation controller (RP-CMAC) with self-generating design, called RPCSGD, is proposed for uncertain nonlinear systems. The proposed controller consists of two parts: one is the parametric CMAC with self-generating design (PCSGD), which is utilized to approximate the ideal controller and the other is the robust controller, which is designed to achieve a specified H^∞ robust tracking performance of the system. The corresponding memory size of the proposed controller can be suitably constructed via the self-generating design. Thus, the useless or untrained memories will not take possession of the space. Besides, the concept of sliding-mode control (SMC) is adopted so that the proposed controller has more robustness against the approximated error and uncertainties. The stability of the system can be guaranteed surely due to the derivations of the adaptive laws of the proposed RPCSGD based on the Lyapunov function. Finally, the proposed controller is applied to the second-order chaotic system and the one-link rigid robotic manipulator. The tracking performance and effectiveness of the proposed controller are verified by simulations of the computer.     

A parametric similarity method: Comparative experiments based on semantically annotated large datasets

We present the parametric method SemSim^p aimed at measuring semantic similarity of digital resources. SemSim^p is based on the notion of information content, and it leverages a reference ontology and taxonomic reasoning, encompassing different approaches for weighting the concepts of the ontology. In particular, weights can be computed by considering either the available digital resources or the structure of the reference ontology of a given domain. SemSim^p is assessed against six representative semantic similarity methods for comparing sets of concepts proposed in the literature, by carrying out an experimentation that includes both a statistical analysis and an expert judgment evaluation. To the purpose of achieving a reliable assessment, we used a real-world large dataset based on the Digital Library of the Association for Computing Machinery (ACM), and a reference ontology derived from the ACM Computing Classification System (ACM-CCS). For each method, we considered two indicators. The first concerns the degree of confidence to identify the similarity among the papers belonging to some special issues selected from the ACM Transactions on Information Systems journal, the second the Pearson correlation with human judgment. The results reveal that one of the configurations of SemSim^p outperforms the other assessed methods. An additional experiment performed in the domain of physics shows that, in general, SemSim^p provides better results than the other similarity methods.     

An efficient solution algorithm for boundary element equations

Boundary element techniques result in the solution of a linear system of equations of the type HU = GQ + B, which can be transformed into a system of equations of the type AX = F. The coefficient matrix A requires the storage of a full matrix on the computer. This storage requirement, of the order of n^*n memory positions (n = number of equations), for a very large n is often considered negative for the boundary element method. Here, two algorithms are presented where the memory requirements to solve the system are only n^*(n - 1)/2 and n^*n/4 respectively. The algorithms do not necessitate any external storage devices nor do they increase the computational efforts.     

Numerical study of laminar natural convection in a complicated cavity heated from top with sinusoidal temperature and cooled from other sides

Steady, laminar natural convection in a two-dimensional enclosure with three flat and one wavy walls is numerically investigated. The top wall is heated with a sinusoidal temperature profile. The other three walls, including the wavy wall, are maintained at constant low temperature. Air is considered as the working fluid. This problem is numerically solved by SIMPLE algorithm with deferred QUICK scheme in non-orthogonal curvilinear co-ordinates. The mesh generation has been done by solving the partial differential equation with grid control functions. Tests are carried out for wave amplitudes 0.0-0.10 in steps of 0.01 and Rayleigh numbers 10⁰-10⁶ while the Prandtl number is kept constant. The number of undulations considered are one, two or three.The effect of the various parameters (Rayleigh number, amplitude of undulation and number of undulations) on the flow pattern and heat transfer has been studied. The heat transfer mode remains conductive up to Ra = 10³. With increase of Ra, the mode of heat transfer changes from conduction to convection. It has been observed that the average Nusselt number remains constant for Ra up to 10³ and then starts changing when Ra is increased further. This led to the further detailed study about the flow behavior in a cavity with undulations. Because of the nature of the imposed boundary conditions, there are two large vortices formed. The left side vortex adjacent to the flat vertical wall always remains single cell and is unaffected by Ra, amplitude and number of undulations. For a single undulation, when Ra is increased to 10⁶, the right side large vortex breaks into two cells for amplitude above a certain value (0.06) giving rise to first saddle point and right side second vortex center. For two and three undulation cases, in addition to these, a second saddle point and right side third vortex appear beyond some amplitude (0.07 and 0.03, respectively).     

Type Destructors

We study a variant of System F_≤ that integrates and generalizes several existing proposals for calculi with “structural typing rules.” To the usual type constructors (→, ×, All, Some, Rec) we add a number of type destructors, each internalizing a useful fact about the subtyping relation. For example, in F_≤ with products every closed subtype of a product S×T must itself be a product S′×T′ with S′<:S and T′<:T. We internalise this observation by introducing type destructors .1 and .2 and postulating an equivalence T=_ηT.1×T.2 whenever T<:U×V (including, for example, when T is a variable). In other words, every subtype of a product type literally is a product type, modulo η-conversion. Adding type destructors provides a clean solution to the problem of polymorphic update without introducing new term formers, new forms of polymorphism, or quantification over type operators. We illustrate this by giving elementary presentations of two well-known encodings of objects, one based on recursive record types and the other based on existential packages. The formulation of type destructors poses some tricky meta-theoretic problems. We discuss two different variants: an “ideal” system where both constructors and destructors appear in general forms, and a more modest system, F_≤^TD, which imposes some restrictions in order to achieve a tractable metatheory. The properties of the latter system are developed in detail.