Abstract- In distant-talking environments, speech recognition performance degrades significantly due to noise and reverberation. Recent work of Michael L. Selzer shows that in microphone array speech recognition, the word error rate can be significantly reduced by adapting the beamformer weights to generate a sequence of features which maximizes the likelihood of the correct hypothesis. In this approach, called Likelihood Maximizing Beamforming algorithm (Limabeam), one of the method to implement this Limabeam is an UnSupervised Limabeam(USL) that can improve recognition performance in any situation of environment. From our investigation for this USL, we could see that because the performance of optimization depends strongly on the transcription output of the first recognition step, the output become unstable and this may lead lower performance. In order to improve recognition performance of USL, some post-filter techniques can be employed to obtain more correct transcription output of the first step. In this work, as a post-filtering technique for first recognition step of USL, we propose to add a Wiener-Filter combined with Feature Weighted Malahanobis Distance to improve recognition performance. We also suggest an alternative way to implement Limabeam algorithm for Hidden Markov Network (HM-Net) speech recognizer for efficient implementation. Speech recognition experiments performed in real distant-talking environment confirm the efficacy of Limabeam algorithm in HM-Net speech recognition system and also confirm the improved performance by the proposed method.

The embedded software has been developed in the forms of various versions that provides similar service based on product family. For increase usefulness of product family, software must has well-structured and reusable properties. Software architecture is important to improve adaptability in model-based development of embedded software mounted onto product family. In this paper, we proposed a technique of ADD(Attribute-Driven Design)-based software architecture design for low power software development. This technique provides a chance to consider the power consumption issue in design phase of software, and makes possible to develop low power embedded software.

In recent, the demand of Unmanned Aerial Vehicles (UAVs) that can autonomous navigation and remote control has been increased in military, civil and commercial field. Particularly, existing researches focused on autonomous navigation method based on vanish point and remote control method based on event processing in indoor environments. However, the existing methods have some problems. For instance, a detected vanish point in intersection point has too much detection errors. In addition, the delay is increased in existing remote control system for processing images in real time. Thus, we propose improved vanish point algorithm by removing detection errors in intersection point. We also develop a remote control system with android platform by separating flying control and image process. Finally, we compare the proposed methods with existing methods to show the improvement of our approaches.

In this research, we report on the development of simulation system for performance verification of BMS(Battery Management System) which is utilized in electric vehicles. In the industrial circles, a manufacturer of BMS typically tests their system with real battery packs. However, it takes a long time to test all functions of BMS. Here, we develop BMU(Battery Managament Unit) as an embedded board, which will be installed in electric vehicle for controlling battery packs. All other environment factors for testing BMU are developed in softwares in order to reduce the term of test. Especially, the proposed system consists of cell simulator and CMU(Cell Management Unit) simulator which simulate real battery cells and control battery cells. These simulators enable the BMU to test more battery cells. In addition, proposed system provides diagnosis program in order to diagnose and monitor the condition of BMS which makes the test of BMS more easily. In order to verify the performance of the developed simulator, we have performed the experiment with real battery packs and our simulator. Through comparing two results of experiments, we verify that developed simulator shows better performance in terms of less amount of testing duration though having high reliability.

This paper presents a method using Maximum-Likelihood Linear Regression (MLLR) adaptation to improve recognition performance of Limabeam algorithm for speech recognition using microphone array. From our investigation on Limabeam algorithm, we can see that the performance of filtering optimization depends strongly on the supporting optimal state sequence and this sequence is created by using Viterbi algorithm trained with HMM model. So we propose an approach using MLLR adaptation for the recognition of speech uttered in a new environment to obtain better optimal state sequence that support for the filtering parameters' optimal step. Experimental results show that the system embedded with MLLR adaptation presents the word correct recognition rate 2% higher than that of original calibrate Limabeam and also present 7% higher than that of Delay and Sum algorithm. The best recognition accuracy of 89.4% is obtained when we use 4 microphones with 5 utterances for adaptation.

In this paper, we propose an implementation of pH reader on Android smart phone by using Eclipse and SDK. After taking a picture of pH spectrum and tested litmus paper at the same time, we save the picture. From the picture, reference values in RGB and YUB of all pH color in the spectrum are obtained and then those of a chosen point by user in the litmus paper are to be compared with. The distances between the reference colors and the litmus paper's color are measured by Euclidean distance in RGB or YUB coordinate. The nearest two pH values are taken to find the weighted average of the result. Series of test show the proposed scheme is feasible and has performance within 10% error.