Nowadays, Unmanned Aerial Vehicle (UAV) is an important technology for military and security application. Various missions can be done using UAV such as surveillance in unknown areas, forestry conservation, and spying enemy territory. Application which is developed in this research has a purpose to simulate condition in war zone for spying the enemy. Platform used in the experiment is Parrot AR. Drone ver.2.0, an mini quadrotor which was developed by Parrot SA. This quadrotor controlled by Robot Operating System (ROS) framework. The quadrotor will search and recognize some objects and locate their location. Many algorithms were used to do the mission. To recognize object Adaboost Classifier and Pinhole Algorithm were used. The result shows that average error for all scenario is only 0.24 meters.

This paper provides a combination of chemotaxic and anemotaxic modeling, known as odor-gated rheotaxis (OGR), to solve real-world odor source localization problems. Throughout the history of trying to mathematically localize an odor source, two common biometric approaches have been used. The first approach, chemotaxis, describes how particles flow according to local concentration gradients within an odor plume. Chemotaxis is the basis for many algorithms, such as particle swarm optimization (PSO). The second approach is anemotaxis, which measures the direction and velocity of a fluid flow, thus navigating "upstream" within a plume to localize its source. Although both chemotaxic and anemotaxic based algorithms are capable of solving overly-simplified odor localization problems, such as dynamic-bit-matching or moving-parabola problems, neither method by itself is adequate to accurately address real life scenarios. In the real world, odor distribution is multi-peaked due to obstacles in the environment. However, by combining the two approaches within a modified PSO-based algorithm, odors within an obstacle-filled environment can be localized and dynamic advection-diffusion problems can be solved. Thus, robots containing this modified particle swarm optimization algorithm (MPSO) can accurately trace an odor to its source

CCTV is one of the tools that can be used to extract the needed traffic Information. Extracted information from image sequences of CCTV can give us real information about the number of passing vehicles and vehicles speed. In this paper we propose a new method in detecting the number of vehicles and vehicle speed measurement in low light conditions. Headlight detection is used in order to identify the existing vehicle. There are few steps in order to extract the information from CCTV. First for vehicle headlight detection, the vehicles are detected with normalized cross-correlation method and centroid-area-difference. The second step is vehicle tracking. Headlight is used to track the movements of the vehicle. The third step is vehicle counting and vehicle speed measurement; pin-hole and euclidean distance methods are used to estimate the vehicle speed. We have compared the vehicle detection algorithm and vehicle counting-speed measurement. The result shows that the normalized cross correlation method has a higher accuracy than area-centroid difference. The pinhole model also is better in estimating vehicle speed compared to euclidean distance.

This work applies a hierarchical transfer learning to implement deep neural network (DNN)-based multilingual text-to-speech (TTS) for low-resource languages. DNN-based system typically requires a large amount of training data. In recent years, while DNN-based TTS has made remarkable results for high-resource languages, it still suffers from a data scarcity problem for low-resource languages. In this paper, we propose a multi-stage transfer learning strategy to train our TTS model for low-resource languages. We make use of a high-resource language and a joint multilingual dataset of low-resource languages. A pre-trained monolingual TTS on the high-resource language is fine-tuned on the low-resource language using the same model architecture. Then, we apply partial network-based transfer learning from the pre-trained monolingual TTS to a multilingual TTS and finally from the pre-trained multilingual TTS to a multilingual with style transfer TTS. Our experiment on Indonesian, Javanese, and Sundanese languages show adequate quality of synthesized speech. The evaluation of our multilingual TTS reaches a mean opinion score (MOS) of 4.35 for Indonesian (ground truth = 4.36). Whereas for Javanese and Sundanese it reaches a MOS of 4.20 (ground truth = 4.38) and 4.28 (ground truth = 4.20), respectively. For parallel style transfer evaluation, our TTS model reaches an F0 frame error (FFE) of 9.08%, 10.13%, and 8.43% for Indonesian, Javanese, and Sundanese, respectively. The results indicate that the proposed strategy can be effectively applied to the low-resource languages target domain. With a small amount of training data, our models are able to learn step by step from a smaller TTS network to larger networks, produce intelligible speech approaching the real human voice, and successfully transfer speaking style from a reference audio.

Adaptive traffic signal control system is needed to avoid traffic congestion that has many disadvantages. This paper presents an adaptive traffic signal control system using camera as an input sensor that providing real-time traffic data. Principal Component Analysis (PCA) is used to analyze and to classify object on video frame for detecting vehicles. Distributed Constraint Satisfaction Problem (DCSP) method determine the duration of each traffic signal, based on counted number of vehicles at each lane. The system is implemented in embedded systems using BeagleBoard™.