A hybrid agent architecture integrating desire,intention and reinforcement learning

This paper presents a hybrid agent architecture that integrates the behaviours of BDI agents, specifically desire and intention, with a neural network based reinforcement learner known as Temporal Difference-Fusion Architecture for Learning and COgNition (TD-FALCON). With the explicit maintenance of goals, the agent performs reinforcement learning with the awareness of its objectives instead of relying on external reinforcement signals. More importantly, the intention module equips the hybrid architecture with deliberative planning capabilities, enabling the agent to purposefully maintain an agenda of actions to perform and reducing the need of constantly sensing the environment. Through reinforcement learning, plans can also be learned and evaluated without the rigidity of user-defined plans as used in traditional BDI systems. For intention and reinforcement learning to work cooperatively, two strategies are presented for combining the intention module and the reactive learning module for decision making in a real time environment. Our case study based on a minefield navigation domain investigates how the desire and intention modules may cooperatively enhance the capability of a pure reinforcement learner. The empirical results show that the hybrid architecture is able to learn plans efficiently and tap both intentional and reactive action execution to yield a robust performance.     

Efficient simulation of hard disk drive operational shock response using model order reduction

The complex structure, coupled mechanical and fluidic energy domains, and inherent nonlinearity of air bearing between slider and disk involved in the hard disk drive (HDD) are normally presented as a large scale problem which will result in very heavy computational costs in terms of intensive computation and time consuming for HDD research communities and industries to carry out the transient dynamic simulation for HDD design verification, performance analysis, and optimization by using the traditional full-order models, such as finite element model (FEM). This paper presents a method of application of model order reduction (MOR) technique to dramatically reduce the computation time for HDD transient shock performance analysis while capturing the behaviors of original problem faithfully. The reduced models are obtained by performing MOR directly to the FEMs through Krylov subspace and Arnoldi algorithm. The transient operational shock response results of the reduced models of a head suspension assembly (HSA) subjected to half-sine shock pulse demonstrate that the reduced models can dramatically reduce total computation by at least three orders and have very good agreement with those simulated from the original large problem by full-order FEM.     

A surface blending approach for displacement features on freeform surfaces

Embedding a number of displacement features into a base surface is common in industrial product design and modeling, where displaced surface regions are blended with the unmodified surface region. The cubic Hermite interpolant is usually adopted for surface blending, in which tangent plane smoothness across the boundary curve is achieved. However, the polynomial degree of the tangent field curve obtained symbolically is considerably higher, and the reduction of the degree of a freeform curve is a non-trivial task. In this work, an approximation surface blending approach is proposed to achieve tangential continuity across the boundary curve. The boundary curve is first offset in the tangent field with the user-specified tolerance, after which it is refined to be compatible with the offset curve for surface blending. Since the boundary curve is offset in a three-dimensional (3D) space, the local self-intersection in the offset curve is addressed in a 2D space by approximately mapping the offset vectors in the respective tangent planes to the parameter space of the base surface. The proposed algorithm is validated using examples, and the normal vector deviation along the boundary curve is investigated.     

Building language resources for a Multi-Engine English-Filipino machine translation system

In this paper, we present the building of various language resources for a multi-engine bi-directional English-Filipino Machine Translation (MT) system. Since linguistics information on Philippine languages are available, but as of yet, the focus has been on theoretical linguistics and little is done on the computational aspects of these languages, attempts are reported here on the manual construction of these language resources such as the grammar, lexicon, morphological information, and the corpora which were literally built from almost non-existent digital forms. Due to the inherent difficulties of manual construction, we also discuss our experiments on various technologies for automatic extraction of these resources to handle the intricacies of the Filipino language, designed with the intention of using them for the MT system. To implement the different MT engines and to ensure the improvement of translation quality, other language tools (such as the morphological analyzer and generator, and the part of speech tagger) were developed.     

Registration using natural features for augmented reality systems

Registration is one of the most difficult problems in augmented reality (AR) systems. In this paper, a simple registration method using natural features based on the projective reconstruction technique is proposed. This method consists of two steps: embedding and rendering. Embedding involves specifying four points to build the world coordinate system on which a virtual object will be superimposed. In rendering, the Kanade-Lucas-Tomasi (KLT) feature tracker is used to track the natural feature correspondences in the live video. The natural features that have been tracked are used to estimate the corresponding projective matrix in the image sequence. Next, the projective reconstruction technique is used to transfer the four specified points to compute the registration matrix for augmentation. This paper also proposes a robust method for estimating the projective matrix, where the natural features that have been tracked are normalized (translation and scaling) and used as the input data. The estimated projective matrix will be used as an initial estimate for a nonlinear optimization method that minimizes the actual residual errors based on the Levenberg-Marquardt (LM) minimization method, thus making the results more robust and stable. The proposed registration method has three major advantages: 1) It is simple, as no predefined fiducials or markers are used for registration for either indoor and outdoor AR applications. 2) It is robust, because it remains effective as long as at least six natural features are tracked during the entire augmentation, and the existence of the corresponding projective matrices in the live video is guaranteed. Meanwhile, the robust method to estimate the projective matrix can obtain stable results even when there are some outliers during the tracking process. 3) Virtual objects can still be superimposed on the specified areas, even if some parts of the areas are occluded during the entire process. Some indoor and outdoor experiments have been conducted to validate the performance of this proposed method.     

Plastic Injection Molding of High-Aspect Ratio Micro-Rods

Technological advancements have led to an increase in demand for fabrication of small and cheap miniature components or parts, especially in the medical and electronic fields. In this research work, micro-rods of varying lengths and diameters were molded using a specially designed tabletop injection molding machine. The main purpose was to investigate whether complete filling of the microcavities was possible and whether small cavity openings will restrict melt flow into the cavities.

Process parameters, such as injection pressure, mold temperature and melt temperature were varied. The micro-rods obtained were analyzed with Scanning Electron Micrographs (SEM). The results showed that injection pressure was the most important parameter for microinjection molding. The use of a vacuum system also yielded better results.     

Quantification of multiple bioagents with wireless, remote-query magnetoelastic microsensors

This paper presents a micromagnetoelastic sensor array for simultaneously monitoring multiple biological agents. Magnetoelastic sensors, made of low-cost amorphous ferromagnetic ribbons, are analogous and complementary to piezoelectric acoustic wave sensors, which track parameters of interest via changes in resonance behavior. Magnetoelastic sensors are excited with magnetic ac fields, and, in turn, they generate magnetic fluxes that can be detected with a sensing coil from a distance. As a result, these sensors are highly attractive, not only due to their small size and low cost, but also because of their passive and wireless nature. Magnetoelastic sensors have been applied for monitoring pressure, temperature, liquid density, and viscosity, fluid How velocity and direction, and with chemical/biological responsive coatings that change mass or elasticity, various biological and chemical agents. In this paper, we report the fabrication and application of a six-sensor array for simultaneous measurement of Escherichia coli O157:H7, staphylococcal enterotoxin B, and ricin. In addition, the sensor array also monitors temperature and pH so the measurements are independent from these two parameters.