The orchestration and management of 5G network slicing (NS) requires cross-domain orchestration across 5G radio access network (RAN), 5G core network (CN), and transport network and also coordination of closed loops of each 5G network segment. We propose a resource orchestration and management architecture for end-to-end (E2E) 5G network slices to automate the flexible and high-performance NS’s management and orchestration. Through developing a prototype of the proposed architecture on an experimental 5G network, the effectiveness of our architecture is proven through prominent use cases certified by ETSI ZSM.

The decomposition of a 5G radio access network (RAN) accelerates inter-operator infrastructure sharing of the corresponding transport network as well as the centralized deployment of a centralized unit (CU) and distributed unit (DU). We propose an architecture that enables inter-operator infrastructure sharing of 5G xHaul as slices and on-demand creation of xHaul network slices dynamically over multiple network operators that are interconnected and orchestrated with the standardized interface of the Open Networking Foundation (ONF) Transport API (TAPI) and corresponding standard from Metro Ethernet Forum (MEF). Through a proof of concept, we validated the implementation of the architecture in three live network domains for 5G xHaul.

Our goal is to estimate athlete 3D pose from monocular TV sports video with a lower cost of collecting training data. To achieve this goal, we utilize a pre-trained deep neural network as a 3D pose estimator for estimating human 3D pose from 2D joint locations of the person in each image. Each image in popular datasets used for training such 3D pose estimator is obtained from a camera whose axis is parallel to the ground. On the other hand, since an image in TV sports video is generally taken from a bird's eye view, joint locations of a human is distorted in the lower part of the image. Therefore, it is not appropriate to give 2D joint locations of the person directly to the pre-trained 3D pose estimator. To resolve this problem, we propose to correct 2D joint locations in an image of TV sports video by a homography transformation that maps the points in the image of TV sports video to the corresponding points in the image taken by the camera that captures training data for the 3D pose estimator. Experimental results show that the proposed method can estimate athlete 3D pose from monocular TV sports video.

Estimating the distance of a target object from a single image is a challenging task since a large variation in the object appearance makes the regression of the distance difficult. In this paper, to tackle such the challenge, we propose 2.5D anchors which provide good candidates of distances, based on a perspective camera model. This candidate is expected to relax the difficulty of the regression model since only the residual from the candidate distance needs to be taken into account. We show the effectiveness of our proposed anchors, by comparing with ordinary regression methods, through experiments with Pascal 3D+ TV monitor dataset and Tsukuba challenge task.

For accurate global self-localization with small memory usage, researches for the compression of the laser-scan data have been actively conducted. Main approaches to the compression are to design feature extractor based on human knowledge regarding the specific environment, e.g., office and hallway. However, in real robot navigation tasks such as a security patrol robot, the robot would be applied to a variety of environments and it is expensive if the users need to tune the design at every environment. To alleviate such problem, we propose to extend the state-of-the-art variational auto-encoder (VAE) by introducing the step-edge detector, which detects non-continuous transition emerged frequently at the laser scan data due to the limitation of distance measurement. With our proposed method, called "laserVAE", the feature extractor of the laser scan is automatically tuned given unknown environments. Through experiments with a real self-localization with 2D laser scan, we demonstrate the effectiveness of the proposed method.

To improve the efficiency of DC-DC converters, this paper focuses on air-core reactors. First, the author deduces the relation between load loss and no-load loss, which originates from ripple current. Next, the author tests a reactor regarding the increase in temperature by changing the waveforms and frequencies under the same cooling and rms current conditions that enable the separation of the DC and AC losses. The results regarding the increase in temperature are different and demonstrate the existence of AC loss in the air-core reactor. It is determined that the AC loss is the eddy-current loss by a finite element analysis. Finally, this paper calculates no-load loss in the DC-DC converter. The AC loss, which consists of eddy-current loss, accounts for more than 90% of the no-load loss. From the results, this paper refers to the importance of the reduction in ripple current in the reactor.

Today, as LSI devices are increasingly more integrated resulting in larger number of package pins, high speed signal lines are more easily coupled to each other. Additionally, since multisite testing is required to reduce the cost of test, the transmission lines connecting these pins become even more concentrated. To measure these LSI devices, a huge number of high-speed transmission lines are required in the test board. Therefore minimizing crosstalk is very important. Furthermore, since the number of channels have reached more than ten thousand in recent evaluation systems, it is strongly required that the crosstalk from all channels be measured. With so many channels, measuring crosstalk with traditional methods will at best be very time-consuming or at worse, completely impractical, as it requires either a large number of pulse generators or physical reconnections depending on measurement type. In this paper, we propose a very fast and low cost crosstalk measurement method, which can acquire results very quickly and with high frequency resolution. Only a single measurement makes it possible to evaluate the crosstalk over a wide frequency band without the repetition of manually measuring at each frequency. This method is highly beneficial to the automatic test equipment (ATE) environment with high density transmission lines on the performance boards or load boards.

This paper shows human skill acquisition systems to control the kite-based tethered flying robot. The kite-based tethered flying robot has been proposed as a flying observation system with long-term activity capability[1]. It is a relatively new system and aimed to complement other information gathering systems using a balloon or an air vehicle. This paper shows some approaches of human operation characteristics acquisition based on fuzzy learning controller, k-nearest neighbor algorithm, and artificial neural network for the kite-based tethered flying robot using human operation data and their validity through computational simulation which we developed[2].

Information from the sky is important for rescue activity in large-scale disaster or dangerous areas. Observation system using a balloon or an airplane has been studied as an information gathering system from the sky. A balloon observation system needs helium gas and relatively long time to be ready. An airplane observation system can be prepared in a short time and its mobility is good. However, a long time flight is difficult because of limited amount of fuel. We have proposed and developed a kite-based observation system that complements activities of balloon and airplane observation systems by short preparation time and long time flight[1]. This research aims at construction of the autonomous flight information gathering system using a tethered flying unit that consists of the kite and the ground tether line control unit with a winding machine. This paper proposes fuzzy controllers for the kite type tethered flying robot inspired by how to fly a kite by a human.

Recently, two-wheeled inverted pendulum mobile robots have been popular. They support human locomotion and/or small goods transportation based on inverted pendulum upright controllers. The conventional inverted pendulum mobile robot controls to follow the fixed desired posture angle and wheel velocity. It is desirable to change the control parameters according to the user intention in order to offer comfortable operability of the robot. This paper proposes a user intention recognition system for a inverted pendulum mobile robot and shows experimental results.

This paper proposes a new global scan matching algorithm using the CIF descriptors and a geometric constraint between keypoints. The CIF descriptor was proposed in our previous work. It is a feature decriptor that is invariant against a congruence transformation. In our previous work, our method was able to perform robust local scan matching using CIF decriptors, but was apt to fail global scan mathching where a large map is used as the reference scan. In this paper, in order to resolve this problem, we propose to use a geometric constraint between keypoints in addtion to the CIF decriptors for the global scan mathching task. Our method can perform global scan matching in a cluttered environment without using an initial alignment. Through experiment in real environment, we confirm the validity of our method by comparing the performance of our method and that of our previous method.

The ability of computing similarities between two data sets is a key for many applications such as video tracking, object recognition, image stitching, 3D modeling and so on. Recently, Lowe has discovered a promissing approach for matching 2D images based on the local invariant feature descriptor called SIFT [1]. We are really inspired by Lowe's method. In this paper, we propose a new local invariant feature descriptor for matching 2D scan data. The proposed feature descriptor is called "CIF", that is a feature which remains unchanged when a congruence transformation is applied. We can perform global scan matching in cluttered environments by matching an input scan with a reference scan based on CIF without any initial alignments. the validity of our method is confirmed by experiments in real environment.

This paper proposes a method for learning the kicking motion of an inverted-pendulum mobile robot from the observation of a human player's demonstration. An inverted-pendulum mobile robot with upper and lower body links observes the human demonstration with a motion capture system and estimates the posture of each human links. The robot maps the links to its own two links and estimates link posture trajectories during the kicking motion. The robot starts learning kicking based on the trajectory parameters for imitation. Through this process, our robot can learn dynamic kicking shown by a human. The mapping gives an important role for successive imitation. A reasonable and feasible procedure of learning from observation for an inverted-pendulum robot is proposed. Learning performance from observation is investigated with a preliminary experiment.

To reduce joule loss due to ripple current in multiphase current reversible chopper, we manufactured reactors, which achieve an optimal electromagnetic-coupling coefficient and we tested the new reactors. As a result, the new reactors whose electromagnetic-coupling coefficient is 0.60 reduces ripple current both on each phase and on the total combined when those operate with differential coupling and shift-phase switching. The loss from ripple current is reduced to 31% when the chopper operates with above mentioned condition. When the chopper operates with equal-phase switching and cumulative coupling and employs new reactors whose electromagnetic-coupling coefficient is 0.97, the joule loss due to ripple current is reduced to 11%.

The need for larger and faster memories has been a constant requirement in the last decades together with keeping memory costs constant or lower. This presents a significant challenge for cost effective memory testing, not only because of the increased data rates but also the pressure to keep memory testing costs down. This paper addresses one of these challenges, which is the development of driver-sharing designs to allow the development of DDR test solutions with a high number of sites. This paper will describe in detail the challenges that high-volume ATE testing of DDR4 presents in regard to driver sharing, allowing the test engineer to better grasp the problems associated with DDR4 high-volume ATE testing.

As Great East Japan Earthquake occurred at 11 March 2011 suffered a heavy loss, Japan is one of the most risky country stroke by an earthquake. Therefore, it is a social mission for enterprises in Japan, especially for telecommunication careers like us, to establish disaster recovery plan against a great earthquake. One way to achieve disaster recovery is locating a backup system geographically away from primary system. Authors have invented D3A (Distributed Data Driven Architecture) that makes thousands of small IA servers cooperated and derives high-performance. We have introduced the technology in commercial large-scale OSS (Operation Support System) to maintain tens of thousands of NE (Network Element) which constitute our telecommunication network. We have already possessed a backup system of the OSS for disaster recovery purpose. However, the backup system leads to great rise in TCO (Total Cost of Ownership) since it also consists of more than 1,000 machines. In this paper, we discuss about an autonomously continuous distributed disaster recovery system that is made up by several locations and every location is considered as primary system equivalently, and outage of one location does not impact on overall service of the system.

This paper addresses development of body mapping from a human demonstrator to an inverted-pendulum mobile robot for imitation. An inverted-pendulum mobile robot with torso and body links learns a dynamic kicking motion shown by a human. The robot observes the human demonstration with a camera, extracts the human region in each of images, maps the region to its own two links, estimates the link posture trajectories, and starts kicking motion learning based on the trajectory parameters for imitation. The mapping parameter gives an important role for successive imitation. A reasonable and feasible procedure of learning from observation for an inverted-pendulum robot and development of the body mapping is proposed and investigated in this paper.

A sensor network covering a large area enables connection to various types of sensors and actuators, but application development becomes complicated because of the uncontrollable behavior of such a large number of devices. We propose the u Tuple Space model for uniform and indirect communication with two extensions to the original tuple space model. The extensions enable efficient range search for multi-dimensional keys and bi-directional communications. Our implementation of our proposed model also integrates load balancing by using a distributed hash table, and the experimental results indicate good scalability in multiple servers. The trial application of gathering and plotting GPS sensor data works well in the field.

Mobile robots operating in indoor spaces, such as office, can benefit from a two-dimensional (2D) map. The map can be used for planning and executing operations such as navigation between different locations in an office. The map might serve as a visual aid to a security operator. This paper presents a method for constructing a 2D map by fusing local images from a special catadioptric sensor taken from several locations in the environment into a single, consistent image.

A method for behavior acquisition that considers embodiment by associating tactile information to visual input is described. An agent that acts in the physical world always suffers constraints derived from embodiment. On the other hand, embodiment plays a very important role in the formation of visual function. Philosophical and clinical medicine findings assert that vision does not function without learning through experiences of haptic motion. In this paper, we discuss the relation between vision, embodiment, and behavior. We develop a method for behavior acquisition through associating vision and tactile sensors. We perform experiments of obstacle avoidance using a computer simulation and a real agent to test the validity of our method.

The conventional reinforcement learning method has problems in applying to real robot tasks, because such method must be able to represent the values in terms of infinitely many states and action pairs. In order to represent an action value function continuously, a function approximation method is usually applied. In our previous work (2000), we pointed out that this type of learning method potentially has a discontinuity problem of optimal actions for a given state. In this paper, we propose a method for estimating where a discontinuity of the optimal action takes place and for refining a state space incrementally. We call this method an continuous valued Q-learning method. To show the validity of our method, we apply the method to a simulated robot.

For future applications, it is important to develop systems using small objects. Such systems must be able to provide uninterrupted services even when some small objects stop. Replicating such objects is one way to do this. However, introducing some type of redundancy into systems generally adds some overhead. Our proposed model reduces this overhead. It is implemented with multi-threaded execution for applications in actual systems. We measured the performance of an implementation for applications connected to databases. The results show that the overhead for ordinary execution and the time required to switch the replicas are both acceptably small. This technique will therefore play an important role in future systems.

In the multi-agent environment, it is important to identify position and orientation of multiple robots for accomplishing a given task in cooperative manner. This paper proposes a method for estimating position and orientation of multiple robots using multiple omnidirectional images based on geometrical constraints. Our method reconstruct not only the relative configuration between robots but also an absolute one using the knowledge of landmarks in the environment. Even if there are some obstacles in the environment, our method can estimate absolute configuration between robots based on the results of self-localization of each robot.

We propose a new architecture for recognizing objects based on a concept of "embodiment" as one of primitive functions for a cognitive robot. We define the term "embodiment" as the size and shape of the agent's body, locomotive ability and its sensor. According to embodiment, an object is represented by reaching action paths, which correspond to a set of sequences of movements taken by the agent for reaching the object. Visual information is used to obtain sensorimotor mapping which represents the relationship between the change of object's appearance and the movement of the agent. On the other hands, tactile information is utilized to evaluate the change of physical condition of the object caused by such movement. By means of this method, the agent can recognize an object without depending on its position and orientation in the environment. The experimental result of computer simulation is shown to validate the method.

In order to keep visual tracking systems with color segmentation technique running in a real environment, an online learning method to update models for adapting them to dynamic changes of surroundings needs to be developed. To deal with this problem, we propose an online visual learning method for color image segmentation and object tracking in a dynamic environment. Our method utilizes a fuzzy ART model which is a kind of neural network for competitive learning. The mechanism of this neural network is suitable for online learning and is different from that of a backpropagation type neural network. In order to use the fuzzy ART model for coder segmentation online, we transform the color signal that the framegrabber used yields to a particular color space called Yr/spl theta/ space. To show the validity of our method, we present some results of experiments using sequences of real images.

An input generalization problem is one of the most important ones in applying reinforcement learning to real robot tasks. To cope with this problem, we propose a self-partitioning state space algorithm which can make non-uniform quantization of the state space. To show that our algorithm has generalization capability, we apply our method to two tasks in which a soccer robot shoots a ball into a goal and prevents a ball from entering a goal. To show the validity of this method, the experimental results for computer simulation and a real robot are shown.

To promote robotic soccer research, we need a low cost and portable robot with some sensors and a communication device. To date, there is no platform for robotic soccer. Therefore, each research must build his own robots or utilize robots which are commercially available. This paper describes how to construct a robot system which includes a lightweight and low-cost mobile robot with visual, tactile sensors, TCP/IP communication device, and portable PC where Linux is running. An example of the developed soccer robot system and preliminary experimental results are also shown.

In this paper, we proposed a method by which a stereo vision-based mobile robot learns to reach a target by detecting and avoiding occlusions. We call the internal representation that describes the learning behavior "stereo sketch". First, an input scene is segmented into homogeneous regions by the enhanced ISODATA algorithm with minimum description length principle in terms of image coordinates and disparity information obtained from the fast stereo matching unit based on the coarse-to-fine control method. Then, in terms of the segmented regions including the target area and their occlusion status identified during the stereo and motion disparity estimation process, we construct a state space for the reinforcement learning method to obtain a target reaching behavior. As a result the robot can avoid obstacles without explicitly describing them. We give the computer simulation results and real robot implementation to show the validity of our method.

Many conventional methods for map generation by mobile robots have tried to reconstruct 3-D geometric representation of the environment, which are time-consuming, error-prone, and necessary to transform the map into the information available for the given task. This paper proposes a method to acquire a statistical map representation robust to sensor noise and directly usable for navigation task. The robot is equipped with a ring of ultrasonic ranging sensors and a collision avoidance behaviour is embedded in it. First, the mobile robot explores in the environment in order to store a set of sequences of sonar data, and the principle component analysis is applied to reduce the dimensionality of the sonar data. As a result, each sequence of sonar data can be described as a score pattern of principal components. Next, these patterns are classified into typical local structures of the environment in order for the robot to discriminate them. Finally, a graph representation of the environment is constructed in which nodes and arcs correspond to these local structures and the transition probabilities between them, respectively. The validity of the method is shown by computer simulations and real robot experiments.

A qualitative approach to quantitatively recovering the shape and pose of a straight homogeneous generalized cylinder (SHGC) based on a weak Lambertian assumption is described. This assumption relaxes the strict cosine law of the Lambertian reflection model. The method does not need to know the lighting condition or surface albedo. The image of the projection of the axis of an SHGC is extracted. The slant angle of the SHGC is estimated using the weak Lambertian assumption along an extremal cross-section curve. The location of the SHGC's 3-D axis on other parallels is located. As a result, the pose (slant and tilt) and the shape (the shape of the cross section, the location of the axis, and the sweeping function) of an SHGC are recovered. Experimental results for both synthesized and real images are shown.>

Weak Lambertian assumption is proposed and used to determine shape and pose of cylindrical objects from a monocular intensity image. The method does not require the knowledge of lighting conditions (light intensity and lighting direction), surface properties, or albedos. Experimental results for both synthesized and real images showing the validity of the method are presented.>

A qualitative approach to quantita- tive recovery of cylindrical shape, pose and illumi- nant cpditiop is described. ''Weask Lambertian gs- sumption is introduced as a qualitative constraint on shading, which can relax the requirements for shape from shading methods in two ways: one is that it does not strictly constrain the property of the perfectly Lambertian surface of objects, and the other is that object surfaces may include spec- ular component in addition to diffused one. The method does not need to know the lighting con- dition (light intensity and lighting direction) or surface albedos. Input scenes include cylindrical objects each of which has a cylindrical surface and planar one as its cross-section. First, an actual lighting condition is transformed into a normal- ized lighting one, and an equation which relates the cross-section contour on the image plane to the shape parameter of the cylindrical object is derived based on the Weak Lambertian Assumption. In the case of scenes including plural cylindrical ob- jects, we can estimate the actual lighting direction and albedos of both diffused and specular com- ponents for each surface. Moreover, we can infer relative configuration of plural objects by making a comparison between an input image and a recon- structed image synthesized by using the estimated actual lighting direction and albedos. Experimen- tal results for both synthesized and real images are shown.