◎Minami Sasaki，Toshiki Ohtsu，Shuichiro Yamada，Hao San，Tatsuji Matsuura，Masao Hotta（Tokyo City University）

This paper presents 0.7V 12bit 3Msps successive approximation register analog-to-digital converter (SAR ADC) in 65nm SOTB CMOS technology. By utilizing substrate voltage control technique of SOTB CMOS, SAR ADC can operate with supply voltage as low as Vdd=0.7V. SPICE simulation results show the feasibility and reliability of proposed technique.

◎Shuichiro Yamada，Toshiki Ohtsu，Minami Sasaki，Hao San，Tatsuji Matsuura，Masao Hotta（Tokyo City University）

This paper presents a low supply voltage non-binary 14bit cyclic analog-to-digital converter (ADC) in 65nm SOTB CMOS technology. By utilizing substrate voltage control technique of SOTB CMOS, cyclic ADC can operate with supply voltage as low as Vdd = 0.7V. SPICE simulation results show the feasibility and reliability of proposed technique to realize a high-resolution ADC at low supply voltage.

◎Jing Wang，Yasuaki Watanabe，Takayuki Sato（Tokyo Metropolitan University）

The purpose of this research is to measure the absolute displacement of a polished surface using a laser Doppler vibrometer. In addition, laser speckle interference, which is a correlation function utilized in a image processing, was detected using a 377-nm ultraviolet laser and ceramic resonator. Experimental results show that the laser Doppler vibrometer and laser speckle interference's correlation functions are shown to be in close agreement.

○Khilda Afifah，Retdian Nicodimus（Shibaura Institute of Technology）

This work evaluates the effect of electrode-skin impedance and DRL circuit to Common-Mode (CM) voltage.Simulation result shows significant effect to CM voltage when DRL circuit is used and the electrode-skin impedance in right leg is increased. Mismatch between the impedance of right and left arm degrades the performance of DRL circuit in suppressing common mode noise.

◎Masahiro Saruta，Nobukazu Takai，Satosi Konno（Gunma University）

Comparators are utilized for ADC parts in IC, and high-speed and low-power consumption comparators are required. However, to realize their characteristics is difficult. Then, we proposed the comparator design by utilizing Q-Learning. To further improve the accuracy of this method, we need to consider corner analysis because circuit characteristics vary by the corner of MOSFET process. Therefore, we propose the decision of an element value by utilizing Q-Learning in consideration of corner analysis. We succeeded in the automatic design of high-speed and low-power consumption comparator while considering of corner analysis in the target circuit.

◎Satoshi Konno，Nobukazu Takai（Gunma University）

A Genetic Algorithm is one of the algorithms which is used in automatic designs of circuit topology, however, automatically designed topology tends to be undecipherable. In this paper, we propose an automatic design method to create a decipherable circuit topology using a Genetic Algorithm. Giving some appropriate constraint can achieve a decipherable circuit topology. From simulation results, we could design a new circuit topology which satisfies given specifications.

◎Shingo Arai，Nobukazu Takai，Yusuke Kubo（Gunma University），Akio Iwabuchi，Soichiro Shimokawa，Yoshitaka Takemoto，Takanaga Yamazaki（Sanken Electric）

In a DC-DC converter, a control circuit decides a stable output voltage, and PID control which consists of 3 pole 2 zero transfer function is widely used. In this paper, we found higher-order transfer function which realizes more stable output using Genetic Algorithm. As an example, we tried to find the higher-order transfer function of the control circuit of the step-down DC-DC converter. From simulation results, we found 5 pole and 4 zero transfer function of control circuit, and we confirmed more high performance of the step-down DC-DC converter was realized.

◎Megumi Chiba，Kazuhiko Hamamoto（Tokai University）

A patient simulator is used to learn a technique of nursing, but it has problems of preparation to learn multiple techniques, cost, and storage space. Therefore, it has been proposed to represent a patient simulator using VR to solve the problems. In previous research of authors’ group, a virtual patient simulator for auscultation has been developed using a head mounted display and a haptic device. The subjective evaluation by academic staffs in dept. of nursing was carried out and the result showed that the simulator was effective. However, some problems remained with the auscultation sounds and the tactile sense. In this research, some improvement of the virtual patient simulator were achieved. As a result of the subjective evaluation, all of the remained problems were solved.

◎Chih Hsiang Chang，Takahiro Asahina，Kenta Shimba，Kiyoshi Kotani，Yasuhiko Jimbo（The University of Tokyo）

Temporal memory within several seconds regulates brain functions, which is expressed as variation in cell activity, implying modification of network dynamic. However, such expression can rely on circuits existing in brain layers. How clusters of single cell activity aggregate as network dynamics also remains unknown. Here, we investigate the questions mentioned above by connecting two dissociated cortical cultures by designed micro-tunnel device. Training stimulation were given first in one subculture, and given later to the other with a delayed time. By setting different time deviation, propagation of spontaneous network burst showed various time span related to designed time deviation, suggesting training success and network balance remodeling to the temporal property given in training.

◎Zugui Peng（Tokyo Institute of Technology），Kenta Shimba（The University of Tokyo），Yoshitaka Miyamoto，Tohru Yagi（Tokyo Institute of Technology）

Gel-supported giant liposomes are spherical vesicles of lipid bilayer which is supported by a hydrogel core. They are useful tools to learn life phenomena, and also expected to be a next-generation biodevice toward industrial application. However, previous formation methods of gel-supported giant liposomes have some limitations since the use of organic solvent and low ionic strength solution. In this paper, we report a method to form gel-supported giant liposomes by gel-assisted approach. Our results show that the gel-supported giant liposome can be formed by our method.

◎Noriyuki Nakazato（Tokyo Denki University），Takeshi Matsumoto，Shinichi Nakasuka（The University of Tokyo），Keita Tanaka（Tokyo Denki University）

We are developing a nano satellite, HATOSAT. In this study, we aimed to develop and prototype antennas for Store & Forward ground station which is one of HATOSAT mission, and to evaluate its characteristics. The antenna which used in this developing is a patch antenna because it can be made smaller and lighter and does not use a mechanical mechanism such as a deployment mechanism. The antenna for the Store & Foward ground station was initially simulated using FR 4, but the antenna gain was 2.5 9dBi, which is less than the target 5dBi. Therefore, it is decided to use the substrate which dielectric loss tangent is lower. Also, we will measure the axial ratio to confirm that the circular polarization is correctly emitted in the future.

◎Ryota Kato，Kenji Suyama（Tokyo Denki University）

The design method of CSD(Canonic Signed Digit) coefficient FIR(Finite Impulse Response) filter using MPGA(Multiple Population Genetic Algorithm) is proposed in this paper. MPGA has multiple populations with different total available nonzero digits. As a result, diversification can be prompted. From design examples, the effectiveness of the method is presented.

◎Yohei Yamauchi，Mitsuyuki Saito，Takahiko Ono（Hiroshima City University）

Automatic driving system requires simple and accurate car model that can accurately represent the state of the car and has a less calculation processing. We have proposed a simple and accurate car model that represents the center of gravity position error due to various non-linear characteristics during turning as a deviation of the front wheel angle based on the linear car model. However, this car model may not be able to sufficiently compensate the longitudinal direction center of gravity position error in situations such as slopes. We propose a simple and accurate car model that included not only the turning direction of the car also the nonlinear characteristics in the longitudinal direction. The effectiveness of the car model is shown in numerical simulation on slope with CarSim.

◎Yuto Maruhashi，Mitsuyuki Saito，Takahiko Ono（Hiroshima City University）

The output feedback control which derives the semi-optimal solution which makes the minimum the cost function which assumes system noise and includes the state and the control input is proposed. The modified output feedback control which assumes system noise and output observation noise and evaluates only the state as a method of deriving the optimal solution is proposed. Considering an actual system, we think that it is not realistic to evaluate the state where it cannot observe. Moreover, although output observation noise may be removable by designing a filter, it is difficult to remove the state observation noise mixed in an internal state. We show the solvability condition of the control problem which assumes system noise and state observation noise and evaluates only the output.

◎Takumi Haida，Mitsuyuki Saito，Takahiko Ono（Hiroshima City University）

The output feedback control proposed until now is added to system noise and when observing an output, the probable output feedback control in consideration of the output observation noise mixed in an observation process is proposed. Although this proposing method assumes the actual system since only state is treated as a cost function There is a problem of being hard to apply to the actual control problem which has a limitation in the control input. Then, system noise and output observation noise is taken into consideration, Let quadratic of output power, and quadratic of control input be a cost function, The control problem which asks for the output feedback gain which minimizes this cost function is proposed. As opposed to such a control problem, we show the solvability conditions.

◎Ryuta Yamazaki，Kazuhiko Hamamoto（Tokai University）

The Ocean Cultural Center will be constructed in 2020 in Shizuoka prefecture. The plan of the center includes an attractive exhibition of the deep-sea environment of Suruga Bay and VR will be applied to realize the exhibition. Especially, the VR contents which enables a visitor to walk around the deep-sea is requested. However, the area of Suruga Bay is very large and the area is dotted with the attractive points. In this paper, a new space presentation method using HMD and omnidirectional camera is proposed for the VR contents of taking a walk in the deep-sea of Suruga Bay. The VR contents can provide a virtual space where a visitor can move from/to an attractive point to/from other attractive point instantaneously and seamlessly.

◎Wen Song，Chandramitasari Widyaning，Shigeru Fujimura（Waseda University）

Electricity consumption forecast plays a significant role in the electric supply management system. Power companies need to maintain a balance between power demand and supply for customers power consumption is always affected by several factors. Our goal is to predict the electricity consumption of the manufacturing company every half an hour the next day. In our work, we proposed a deep learning neural network model based on the Long Short-Term Memory(LSTM). Our proposed method performs several experiments in actual time series data of the manufacturing company's power consumption. The experiment result shows that the proposed method outperforms the previous research of LSTM-FFNN and Moving Average(MA) based on the loss of Root Mean Squared Error(RMSE) score.

◎Kenta Uehara，Yoshiki Hatakenaka，Shigeharu Shimamura，Seiichi Koakutsu（Chiba University）

Nowadays, plant factories are attracting worldwide attention as a technology for stably producing crops. One of the problems of plant factory is tipburn, which is a physiological disorder of crops. If tipburn occurs, tipburn leaves are trimmed by hand, or that lettuce is removed from products. Identifying tipburn by eye observation and trimming tipburn leaves by hand requires much labor and cost. In this study, we aim to perform and compare discrimination of tipburn occurrence and tipburn non-occurrence in lettuce cultivated at plant factories using convolutional neural networks with both multiple models and optimization methods. As a result of the experiment, it is confirmed that discrimination can be performed with high accuracy.

◎Ryoma Watanuki，Tadashi Horiuchi，Toshiyuki Aodai（National Institute of Technology, Matsue College）

Deep Q-network (DQN) is one of the most famous methods of deep reinforcement learning. DQN approximates the action-value function using convolutional neural network (CNN) and updates the weights of CNN using Q-learning framework. We have already realized that a single mobile robot acquired behavior to avoid walls and obstacles by using DQN. In this research, we apply deep reinforcement learning to multi-robot environment and we aim to realize that multiple mobile robots acquire behavior to avoid walls and other robots using distance sensors and images of camera mounted on them. This task in multi-robot environment is more difficult than the task in a single robot environment, due to the increase of the state space and the influence by the actions of other robots.

◎Xuanzhi Jin（Zhejiang University），Shigeru Fujimura（Waseda University），Yiping Feng，Gang Rong，Zhengguang Wu，Lei Xie（Zhejiang University）

In a production system, scheduling is important because it can reduce the production cost, which includes both the time cost and manufacturing cost. Thus, many researchers focus on such objectives usually ignores the maintenance of the machines as a part of the producing process. However, the reliability of the machines goes down during the process. So as processing time passes by, maintenance is necessary for the system. This paper discusses different maintenance strategies in a variety of processing scenarios and proposes a new maintenance strategy based on the machine reliability to keep a balance between the cost of production and reasonable reliability.

◎Tatsuya Sugiyama（Kobe University），Kazutoshi Sakakibara，Masaki Nakamura（Toyama Prefectural University），Hisashi Tamaki（Kobe University）

We consider the autonomous vehicle systems where a fleet of autonomous vehicles serves requests of residents’ commute like taxies.In order to realize safe and efficient, not only control systems for individual vehicles but also those for a group of vehicles are essential. Moreover considering the possible presence of traffic jam or deadlocks is also important. In this research, we propose a simulation model for the service which transfers passengers by a fleet of autonomous vehicles.the proposed model consist of vehicle assignment for each request and the vehicle movement control. It can incorporate various techniques and evaluate their performance under realistic situations. We examine the potential of the proposed model through some numerical experiments.

◎Takuto Matsuura，Kazuhiko Hamamoto，Kiyoko Tanaka（Tokai University）

As globalization has been increased recently, intercultural communicative competence is becoming more important for students. However, all of students is not interested in it and they are not confident in their intercultural communication abilities. This is caused by their individual differences and growth environment. Therefore, they can’t proactively communicate with people from other cultures. It is necessary to solve this problem for the students who experience such difficulties with intercultural communication. This study proposes to use VR as a method to solve this problem. A self-learning VR system for students to learn intercultural communication is developed.

○Andrea Porco，Koki Oshiro，Dongshik Kang，Morikazu Nakamura（University of the Ryukyus）

Human emotions estimation became a challenge for the world wide engineers community and the connection between arts, such as the music, and the human emotions, give us an start point to reveal strong patterns still unexplored. In this paper, we propose an estimation method of human emotions from EEG signals taken directly from a Brain Computer Interface (BCI) while listening music. We evaluate valence and arousal over brain waves measured in consecutive segments to map into 4 emotions. The segments evaluated depend on its correlation with music and the possible emotional changes generated from one segment respecting to the preceding. The results show that depend on selected music emotional changes were detected intra segments while a change on selected music could remain them unchanged.

◎Hongran Wang，Kenichi Tamura，Junichi Tsuchiya，Keiichiro Yasuda（Tokyo Metropolitan University）

In this paper, we propose a new Firefly Algorithm (FA) based on superior relations, which can search for the superior solution set by using the user's quantitative desire level as a search strategy. We subsequently analyze the superior solution set search problem and FA, and we discuss its properties together with a similar problem setting. We propose an FA based on superior relations as a new optimization technique for the superior solution set problem based on these analyses. Numerical experiments are then conducted using the superior solution set search problem, and the usefulness of the proposed method is demonstrated while comparing the performance of the proposed method with the conventional FA.

◎Xuyuan Li，Kenichi Tamura，Junichi Tsuchiya，Keiichiro Yasuda（Tokyo Metropolitan University）

In this paper, we introduce a new expanded search mechanism, which is paying attention to updating the group of solutions searched by a same point rather than just a solution. Using this mechanism in multi-point searching, every search point can form a nonoverlapping searched area in solution space of neighborhood relationship, which can solve issues when using common search mechanism. Basing on the high affinities to “Hierarchical Interpretation of Solution Space”, which is also proposed by our research group, we utilize the expanded search mechanism to construct a search tactic among macroscopic movement. And basing on this assembly, we propose a new combinatorial optimization method to verify the basic performance of the expanded search mechanism.

◎Kaijie Yang，Wen Song，Shigeru Fujimura（Waseda University）

Accurate short-term traffic flow prediction is an integral part of Intelligent Transportation System (ITS). There are many time series forecasting methods applied to traffic flow prediction. However, many methods only focus on the time series data of traffic flow at a single observation position without considering the spatial characteristics of traffic flow. In order to solve this problem, this paper proposes a method with up-and-downstream spatial-temporal relationship based on LSTM. The results show that the proposed method can combine temporal information and spatial information of traffic flow to improve the accuracy of prediction.