The target of this work is to optically interconnect a semiconductor laser and a single mode fiber (SMF) through a simple Si bench technology using buried waveguide devices. This technology is suitable for applications such as optical transceivers and add-and-drop multiplexers. Three major components, namely, planarized laser diode, buried waveguide, and SMF are hybrid integrated on the Si bench. The ridge-type laser was planarized by BCB etch-back process, and was flip-chip mounted on the Si bench. On the other hand, the sol-gel buried waveguide was passively aligned to SMF using V-groove and U-groove techniques. Miss-alignment loss as low as 1 dB can be obtained.

The objective of this thesis is to integrate the optical waveguide and single mode fiber in a passive alignment way on a silicon bench. This technique can reduce the complexity of packaging the individual components and increase yield of the module in order to achieve the goal of the mass production. In this module, buried waveguide structure was used for light guidance. A 1.31µm semiconductor laser was used as the input light source. Light signal launched by semiconductor laser is transferred through the buried waveguide into the single mode fiber. This module structure is consisted of two major parts, namely, the buried waveguide and the silicon bench. Buried optical Waveguide uses SO2 as the bottom cladding. Conventional photolithography procedures and etching technique were used to form a trench on the SiO2 cladding. The waveguide core was fabricated by coating the organic-inorganic hybrid materials into the trench. Finally, an organic-inorganic hybrid materials with a refractive index smaller than that of the core is used as the top cladding. The silicon benches were obtained by etching V-groove and saw-cutting U-groove on the silicon substrates for fixing the fiber. The patterning of buried waveguide and silicon V-groove were fabricated by a single optical mask procedure. Therefore accurate alignment between the waveguide and the single mode fiber can be obtained.

A system of Chinese resume by speech construction is developed by the use of a novel segmentation mechanism and the classical Hidden Markov Model. The recognition system is based on both mono-syllable HMM's and speech-text alignment schemes. Experimental results indicate that the amount of training materials used for feature extraction can be greatly reduced, and the text content of the recorded speech training data can be different from those of the recognition tasks as well. Each phrase in the resume can be identified within one second, that is approximately the same as the graduate did last year. Furthermore, the user interface of the resume system has been redesigned and polished by the GTK toolkit in order to enable event-driven X-window operations.

The aim of this thesis is to design a controller for an unmanned helicopter to perform target engagement. This mission is briefly defined as / the helicopter flies to a firing point under the commands of a trajectory controller, and then it is aligned to the target with attitude control. After weapon firing, the helicopter initiates a return maneuver under again the commands of the trajectory controller. This mission where the continuous systems and discrete guidance decisions are to be executed in coherence can be studied as a hybrid control problem. One hybrid control approach which is used in this study is the representation based on two motion primitives: trim trajectories and maneuvers. To obtain the desired trim trajectories and the maneuvers, a dynamic inversion based controller is developed. The controller has two loops: the inner loop which controls the helicopter attitudes and the outer loop which controls the helicopter trajectory. A guidance algorithm is developed which enables the controller to switch from the inner loop to the outer loop or vice versa. Simulations are generated to test the controller performance.

Background: The development of bioinformatics databases, algorithms, and tools throughout the last years has lead to a highly distributedworld of bioinformatics services. Without adequatemanagement and development support, in silico researchers are hardly able to exploit the potential of building complex, specialized analysis processes from these services. The Semantic Web aims at thoroughly equipping individual data and services with machine-processable meta-information, while workflow systems support the construction of service compositions. However, even in this combination, in silico researchers currently would have to deal manually with the service interfaces, the adequacy of the semantic annotations, type incompatibilities, and the consistency of service compositions. Results: In this paper, we demonstrate by means of two examples how Semantic Web technology together with an adequate domain modelling frees in silico researchers from dealing with interfaces, types, and inconsistencies. In Bio-jETI, bioinformatics services can be graphically combined to complex services without worrying about details of their interfaces or about type mismatches of the composition. These issues are taken care of at the semantic level by Bio-jETI’s model checking and synthesis features. Whenever possible, they automatically resolve type mismatches in the considered service setting. Otherwise, they graphically indicate impossible/incorrect service combinations. In the latter case, the workflow developermay either modify his service composition using semantically similar services, or ask for help in developing the missing mediator that correctly bridges the detected type gap. Newly developed mediators should then be adequately annotated semantically, and added to the service library for later reuse in similar situations. Conclusion: We show the power of semantic annotations in an adequately modelled and semantically enabled domain setting. Using model checking and synthesis methods, users may orchestrate complex processes from a wealth of heterogeneous services without worrying about interfaces and (type) consistency. The success of this method strongly depends on a careful semantic annotation of the provided services and on its consequent exploitation for analysis, validation, and synthesis. We are convinced that these annotations will become standard, as they will become preconditions for the success and widespread use of (preferred) services in the Semantic Web

Aligning and fixing the fiber and laser device is an important work in butterfly laser module packaging. Assembling these two component by laser welding is to achieve reliable and stable weld joint. However, during the welding process, rapid solidification of the welded region and the associated material shrinkage causes the fiber position moved. The relative movement between fiber and laser is called post-weld-shift(PWS). A few micrometers PWS makes coupled power lost. Therefore, minimizing the PWS between fiber and laser is a key research topic in butterfly laser module packaging. We can correct the PWS minimal by fiber shift inspect. But it has a difficult of space limit in butterfly laser module packaging. In this study, a video camera with image acquisition system was used to measure the PWS. We also used a mirror image to solve the problem of space limited. The PWS inspection result has matched our simulation. This method can successfully inspect the PWS in butterfly laser module.

In the present studies we investigated the effects of mechanical rubbing on the surface characteristics of polymer liquid crystal alignment layers. Contact angles of water droplets in contacted with the rubbed polyimide were measured using a surface tension meter, and the surface free energy of the polymer thin films were evaluated. We found that the contact angle of water and surface free energy on rubbed polyimide is anisotropic, and rubbing caused decrease in surface free energy and wettability of the polyimide surface. It was also seen that the contact angle hysteresis and the surface free energy measured in the direction parallel to the rubbing direction is smaller than that in the direction anti-parallel to the rubbing direction. We found that when the pile impression of the velvet fibers is 0.3 mm and the cumulative number of rub is 3 times, the contact angle hysteresis in direction parallel or anti-parallel to the rubbing direction will to be close. Be suitable choosing rubbing conditions, the SSFLCs without zigzag defects was produced. The pretilt angle and the response time of liquid crystal increased with the cumulative number.

The liquid-crystal display because matches to the method is different , common has the TN monitor , the STN monitor , and the MVA monitor . Because these match to the method difference , therefore liquid crystal molecule pretilt angle is also different . Present liquid crystal molecule pretilt angle choice way includes : (1) Chooses the different polyimide material , after specific rubded intensity , but produces specific pretilt angle. (2) Uses low pretilt angle the polyimide material , mixes high pretilt angle according to the different proportion the polyimide material , but achieved accent of the pretilt angle changes . But in this research proposed double alignment layers , takes the first floor by homotropic layers , uses homogenous layers does for the upper formation , the surface free energy of the alignment layer can be easily controlled by adjusting the thickness of the top polyimide layer, similarly may achieve the accent to change liquid crystal molecule pretilt angle the effect . And may compensate the liquid crystal display to be insufficient because of rubbed polyimide , but has the question which dark condition is exposed , has a higher contrast gradient . Because this experiment scope accent changes the effect in liquid crystal molecule higher pretilt angle good , also kneading board various regions liquid crystal molecule pretilt the angle uniformity is higher , therefore may effectively apply on various types homotropic liquid crystal display component , for example: MVA , VA-STN , DSTN liquid-crystal display .

Multi-domain Vertical Alignment Flexible Liquid Crystal Display based on photolithography and replica-Molding method has been demonstrated. In order to maintain a uniform cell gap between flexible substrate,the microstructures were fabricated with polydimethylsiloxane (PDMS) material by replica-molding method. The microstructures master were designed and fabricated using a photosensitive resin (SU-8) by photolithography. The microstructures of pixel-encapsulated walls enhance the mechanical strength to prevent the liquid crystal molecules flow in the bend state deformations. Besides, the elastomeric material, PDMS, provide weak surface energy and induce vertical alignment for liquid crystal spontanelusly without any surface treatment. The microstructure protrusions made by PDMS can provide multi-domain vertical alignment (MVA) effect with wide viewing angle and high contrast ratio. Therefore, this method could be implemented for achieving multi-domain vertical alignment on a flexible LCD applications. The flexible LCD have great stability reproducibility, durability and good electro-optical performances.

This work discusses the effect of photo-alignment in a liquid-crystal film that is doped with nanoparticles and azo-dye. The vertical alignment that is induced by nanoparticles of polyhedral oligomeric silsesquioxanes can be changed into homogeneous alignment by the absorption of photo-excited azo-dye. Both electro-optical and surface properties are analyzed to confirm this phenomenon. An electrically tunable polarization-dependent LC phase grating is produced using the method. By this method, we also present a transflective liquid crystal display with a single cell-gap in a dual-alignment configuration. This dual structure was fabricated via a single-step laser-light exposure through a photomask. The vertical alignment induced by nanoparticles of polyhedral oligomeric silsesquioxanes can be changed into a homogeneous alignment via absorption of photo-excited azo dye. The LC molecules at the transmissive and reflective pixels are homogeneous and hybrid alignment, respectively, at which optimal phase retardations of half and quarter wavelengths are achieved. Both nanoparticle- and azo-dye-induced vertical and homogeneous alignments are non-contact alignment processes and have potential for practical application.