Search

Recent developments in conversational technologies have attracted researchers to study their applications in recommending items through conversations. It is considered that physical robots, rather than virtual ones, are effective in situations in which robots talk about items near participants. However, in real situations, robots may be required to recommend items that are present but invisible in the scene of the communication. In this study, we conducted a field experiment in a shopping mall to investigate the effects of robot embodiment on recommendation tasks. The robots recommended a dish after talking to participants about their food preferences. We developed a conversational recommendation system and implemented it using physical and virtual robots. The field experiment was conducted in a shopping mall; the visitors were encouraged to participate. The experiment lasted a total of 99 hours (9 hours per day for 11 days) inside and in front of a food court. Although no significant difference in the behavioral aspect was confirmed, the results obtained from 272 conversations suggested that having physical bodies enhanced the satisfaction and agreement with the robots' recommendations.

Robots are required to be significantly compliant and versatile to work in unstructured environments. In a number of studies, robots have positively exploited the environments during interactions and completed tasks from a morphological viewpoint. Modular robots can help realize real-world adaptive robots. Researchers have been investigating the actuation, coupling, and communication mechanisms among these robots to realize versatility. However, the diverse force transmission among modules needs to be further studied to achieve the adaptive whole-body dynamics of a robot. In this study, we fabricated a modular robot and proposed the realization of force transmission on this robot, by constructing fluid transferable network systems on the actuation modules. By exploiting the physical property variations of the modular robot, our experimental results prove that the robot’s motion can be changed by switching the connection pattern of the system.

In this paper, we present an electrically driven childlike android named ibuki equipped with a wheeled mobility unit that enables it to move in a real environment. Since the unit includes a vertical oscillation mechanism, the android can replicate the movements of the human center of mass and can express human-like upper-body movements even when moving by wheels. Moreover, providing 46 degrees of freedom enables it to perform various human-like physical expressions. The development of ibuki, as well as the implementation and testing of several functions, is described. Finally, we discuss the potential advantages and future research direction of a childlike mobile android.

Computer networks are facing serious threats from the emergence of sophisticated new DGA bots. These DGA bots have their own dictionary, from which they concatenate words to dynamically generate domain names that are difficult to distinguish from human-generated domain names. In this letter, we propose an approach for identifying the callback communications of DGA bots based on relations among the words that constitute the character string of each domain name. Our evaluation indicates high performance, with a recall of 0.9977 and a precision of 0.9869.

The seismic isolation mechanism of the core-suspended isolation (CSI) system comprises a double layer of inclined rubber bearings installed on top of a reinforced core structure. A multilevel structure is then suspended from a hat truss or umbrella girder constructed on the seismic isolation mechanism. The first building to use the CSI system was the Safety and Security Center in Tokyo, Japan, whose structural health monitoring system has detected and recorded 231 earthquake motions since 2006, including the 2011 Tohoku Pacific Earthquake (2011-TPE). The present study estimates the dynamic characteristics of this CSI-equipped building in earthquakes, and the effects of the CSI system are revealed via the observed earthquake records. The temporal changes of the fundamental period and damping factor are estimated from the 2011-TPE; the fundamental period increases with the deformation of the isolation level, whereas the fundamental damping factor is only related weakly to that deformation. The 231 observed earthquake records reveal that the CSI system performs seismic isolation by reducing the response acceleration of the suspended structure by roughly a half for peak ground accelerations exceeding 30 cm/s2. Daily microtremor observations are used to diagnose earthquake damage; the fundamental frequencies in each direction remain almost constant and were not changed by the 2011-TPE.

This research develops a conversational robot that stimulates users’ dialogue satisfaction and motivation in non-task-oriented dialogues that include opinion and/or preference exchanges. One way to improve user satisfaction and motivation is by demonstrating the robot’s ability to understand user opinions. In this paper, we explore a method that efficiently obtains the concept of user preferences: likes and dislikes. The concept is acquired by complementing a small amount of user preference data observed in dialogues. As a method for efficient collection, we propose a dialogue strategy that creates utterances with the largest expected complementation. Our experimental results with a female-type android robot suggest that the proposed strategy efficiently obtained user preferences and enhanced dialogue satisfaction. In addition, the strength of user motivation (i.e., long-term willingness to communicate with the android) is only positively correlated with the android’s willingness to understand. Our results not only show the effectiveness of our proposed strategy but also suggest a design theory for dialogue robots to stimulate dialogue motivation, although the current results are derived only from a female-type android.

Modelling the cognitive process is a challenging task. Contextual conditions and the scope of the options are critical factors that influence human decisions. We propose and formulate an attention-based network to model the various choices made by humans based on the various factors that predict the possible option in each scope. To evaluate our proposed method, we conducted a user choice experiment in which a user chose an option from among limited choices. Our results showed that our model successfully extracted the hidden context on the attention layer and even outperformed the chance level in terms of prediction accuracy.

In diverse situations, humans produce natural and adaptable bipedal locomotion by cooperatively manipulating the interactions among the different parts of their bodies and the environment. Therefore, to realize a robot with adaptable behavior, it should be enabled to adjust its morphology accordingly in response to environmental changes. From this perspective, this study introduces the development of a bipedal robot with adaptive morphology. By implementing an actuator network system (ANS), the robot is able to manipulate the physical characteristics of its legs and the way they interact with each other. Two experiments have been conducted: main and supplementary experiments. The main experiment examined how effective is adjusting the robot’s morphology on changing the robot’s behavior. The experiment was conducted on different ground materials and under different connection patterns between the robot’s legs. During the experiment, the robot’s behavior was evaluated in reference to four aspects: walking style, stability, speed, and moving direction. The supplementary experiment took the results of the main experiment and used it to improve the robot’s behavior during locomotion. The robot was enabled to automatically switch between the different connection patterns of the ANS, which in turn changed the interaction between the robot’s legs and generated a more suitable dynamics for the surrounding environment.

Some of the most serious threats to network security involve malware. One common way to detect malware-infected machines in a network is by monitoring communications based on blacklists. However, such detection is problematic because (1) no blacklist is completely reliable, and (2) blacklists do not provide the sufficient evidence to allow administrators to determine the validity and accuracy of the detection results. In this paper, we propose a malicious DNS query clustering approach for blacklist-based detection. Unlike conventional classification, our cause-based classification can efficiently analyze malware communications, allowing infected machines in the network to be addressed swiftly.

Some of the most serious security threats facing computer networks involve malware. To prevent this threat, administrators need to swiftly remove the infected machines from their networks. One common way to detect infected machines in a network is by monitoring communications based on blacklists. However, detection using this method has the following two problems: no blacklist is completely reliable, and blacklists do not provide sufficient evidence to allow administrators to determine the validity and accuracy of the detection results. Therefore, simply matching communications with blacklist entries is insufficient, and administrators should pursue their detection causes by investigating the communications themselves. In this paper, we propose an approach for classifying malicious DNS queries detected through blacklists by their causes. This approach is motivated by the following observation: a malware communication is divided into several transactions, each of which generates queries related to the malware; thus, surrounding queries that occur before and after a malicious query detected through blacklists help in estimating the cause of the malicious query. Our cause-based classification drastically reduces the number of malicious queries to be investigated because the investigation scope is limited to only representative queries in the classification results. In experiments, we have confirmed that our approach could group 388 malicious queries into 3 clusters, each consisting of queries with a common cause. These results indicate that administrators can briefly pursue all the causes by investigating only representative queries of each cluster, and thereby swiftly address the problem of infected machines in the network.

Computer networks are facing serious threats from the emergence of malware with sophisticated DGAs (Domain Generation Algorithms). This type of DGA malware dynamically generates domain names by concatenating words from dictionaries for evading detection. In this paper, we propose an approach for identifying the callback communications of such dictionary-based DGA malware by analyzing their domain names at the word level. This approach is based on the following observations: These malware families use their own dictionaries and algorithms to generate domain names, and accordingly, the word usages of malware-generated domains are distinctly different from those of human-generated domains. Our evaluation indicates that the proposed approach is capable of achieving accuracy, recall, and precision as high as 0.9989, 0.9977, and 0.9869, respectively, when used with labeled datasets. We also clarify the functional differences between our approach and other published methods via qualitative comparisons. Taken together, these results suggest that malware-infected machines can be identified and removed from networks using DNS queries for detected malicious domain names as triggers. Our approach contributes to dramatically improving network security by providing a technique to address various types of malware encroachment.

Recently, considerable research has focused on personal assistant robots, and robots capable of rich human-like communication are expected. Among humans, non-verbal elements contribute to effective and dynamic communication. However, people use a wide range of diverse gestures, and a robot capable of expressing various human gestures has not been realized. In this study, we address human behavior modeling during interaction using a deep generative model. In the proposed method, to consider interaction motion, three factors, i.e., interaction intensity, time evolution, and time resolution, are embedded in the network structure. Subjective evaluation results suggest that the proposed method can generate high-quality human motions.

The aim of this workshop is to discuss autonomous dialogue technologies in "Symbiotic Human-Robot Interaction." In order to inspire participants and encourage the discussion, we will introduce our research activities and have invited talks by experts in human-robot interaction and dialogue systems. We also publicly invite poster presentations to share state-of-the-art technologies in this research area. Finally, we clarify the key challenges required to develop companions living together with us through the panel discussion.

Androids--humanoids with a human-like appearance are studied in limited communities but draw considerable attention. Considering an android integrates numbers of actuators, sensors and devices into a system, a standardized framework is beneficial to easy-move and extension of potential applications. In this report, we developed a ROS-based software framework for androids. With the framework, intuitively generating, blending and reusing the android motion is aimed to be realized. Some software applications are developed under this framework and they are implemented on a child-like android "ibuki''. Several android behaviors are created by implementing this framework on ibuki.

In this report, we propose a framework for generating long-term human-like motion based on a deep generative model. Thanks to the network structure, the proposed method allows us generating seem- less long-term motions while the model is trained by 4 seconds long short motion samples. The computer graphics of generated motions seem to be reproduced scenes where a pair of persons talking to each other.

This paper presents a subjective evaluation of the emotions of a wheeled mobile humanoid robot expressing emotions during movement by replicating human gait-induced upper body motion. For this purpose, we proposed the robot equipped with a vertical oscillation mechanism that generates such motion by focusing on human center-of-mass trajectory. In the experiment, participants watched videos of the robot's different emotional gait-induced upper body motions, and assess the type of emotion shown, and their confidence level in their answer. We calculated the emotion recognition rate and the average confidence level of their answers. As a result, we found that participants gave higher confidence levels in their assessment for the robot's emotional movement with vertical oscillation than one without it.

Realizing adaptability to environmental changes requires the robot’s body to change according to the environment. From this perspective, several studies have considered variable compliant actuation to change the physical characteristics of the robot as it interacts with the environment. Robots with this ability show a variety of efficient stable motions during contact with the environment. However, having locally variable compliant parts with independent dynamics under the same body may be insufficient for achieving adaptability in diverse environments. To extend the idea of variable compliant actuation, from being used locally to the level of the entire body’s dynamics, in this study, we use an actuator network system (ANS). As an approximation of bipedal robots, we developed an eight-legged rimless wheel robot with an ANS that allows energy transfer among limbs through passive interactions among mutually interconnected air cylinders mounted on the robot’s legs. We experimentally demonstrated that by using the ANS to realize energy transfer among limbs, the robot performance improved.

Bots are some of the most serious threats. Although the administrators must remove bots that reside in their networks, many bots have domain generation algorithms (DGAs) to avoid detection. A DGA is a technique for hiding the callbacks of bots by frequently changing the domain name of a command-and-control server. In this letter, we propose an approach for detecting the callbacks based on the randomness of domain names. Our evaluation indicates high performance, with a recall of 0.9960 and a precision of 0.9029. By enabling one to swiftly address various bots, our approach contributes to dramatically improving network security.

Android robots: humanoids with realistic human appearance can serve as a telecommunication medium (teleoperated humanoids). We have anticipated that if teleoperated humanoids acquired with not only their human appearances, but also an ability of to realize human-like head motions, the telecommunication and the system operability through androids would greatly improve. To realize such human-like head motions of teleoperated humanoids, we referred to the human cervical spine mechanism, and accordingly aimed to verify how the cervical structure of teleoperated humanoids would affect the operability in our remote-controlling system. First, we developed a multiple joint cervical structure for our teleoperated humanoid, and then, we evaluated of if the multiple joint cervical structure realized human-like head motions and improved its system operability through experiments from the aspects of its eyesight targeting performance and subjective impressions. Finally, we concluded that our proposed multiple joint cervical structure of the teleoperated humanoid enabled to increase the system operability in comparison with the prototype system: with either a top joint cervical structure or a bottom joint one. We anticipate that this work will contribute to the future development and implementation of telecommunication by humanoids.

Humanoids are being applied into society gradually in the aspect of human---robot interaction recently. The human spine plays an important role when performing natural human upper body postures. However, most of?the humanoids only show tense body postures due to the limitations of their simple mechanical structures. We investigated that the human natural spine postures can be imitated by serially connected rigid links with a few joints by the analysis of human spine motion. In this report, we proposed a robotic spine composed of parallelogram actuation modules, and built a prototype. We also investigated the natural spinal postural appearances which are realized by the prototype.

In this paper, we propose a novel energy-efficient architecture for software-defined data center infrastructures. In our proposed data center architecture, we include an exhaust heat reuse system that utilizes high-temperature exhaust heat from servers in conditioning humidity and air temperature of office space near the data center. To obtain high-temperature exhaust heat, equipment such as server racks and air conditioners are deployed in tandem so that the aisles are divided into three types: cold, hot, and super-hot. In this paper, to investigate the fundamental characteristics of our proposed data center architecture, we consider various types of data center models and conduct numerical simulations that use results obtained by experiments at an actual data center. Through simulation, we show that the total power consumption by a data center with our proposed architecture is 27 percent lower than that by data center with a conventional architecture. In addition, it is also shown that the proposed tandem equipment arrangement is suitable for obtaining high-temperature exhaust heat and decreasing the total power consumption significantly under a wider range of conditions than in the conventional equipment arrangement.

There are many environment control systems such as air conditioning systems for providing suitable environment to users. However, there are significant individual differences in the sense of suitable environments among users. Hence, a recommendation system indicating personally suitable environments for each user is extremely important. For this purpose, we are now developing an interactive human–machine interface system that dynamically recommends a suitable room environment for users. In this study, we investigated how the anthropomorphic appearance of the recommending system affected the satisfaction of the users. The results showed that the intellectual impression of the appearance was more important than the emotional impression when using the recommendation system.

Humans can infer the body shapes, actions, and emotions of an animal by observing a pattern of moving white dots. This phenomenon is called “biological motion.” However, at times, humans feel perceive animacy even via motion patterns comprising more simple geometric shapes, for example, a circle and a triangle. In this study, we attempted to create generative models of biological motion patterns of up-down circular motion with emotional expressions. We collected motion patterns created by naive participants and attempted to devise generative models that could generate biological motion with emotional expressions based on the gathered data. Our result implied that generative models of biological motion with emotional expressions may be acquired from the collective creations of many individuals.

Recent advances in high performance computing have allowed sampling-based motion planning methods to be successfully applied to practical robot control problems. In such methods, a graph representing the local connectivity among states is constructed using a mathematical model of the controlled target. The motion is planned using this graph. However, it is difficult to obtain an appropriate mathematical model in advance when the behavior of the robot is affected by unanticipated factors. Therefore, it is crucial to be able to build a mathematical model from the motion data gathered by monitoring the robot in operation. However, when these data are sparse, uncertainty may be introduced into the model. To deal with this uncertainty, we propose a motion planning method using Gaussian process regression as a mathematical model. Experimental results show that satisfactory robot motion can be achieved using limited data.

Developing self-organised swarm systems capable of adapting to environmental changes as well as to dynamic situations is a complex challenge. An efficient labour division model, with the ability to regulate the distribution of work among swarm robots, is an important element of this kind of system. This paper extends the popular response threshold model and proposes a new adaptive response threshold model (ARTM). Experiments were carried out in simulation and in real-robot scenarios with the aim of studying the performance of this new adaptive model. Results presented in this paper verify that the extended approach improves on the adaptability of previous systems. For example, by reducing collision duration among robots in foraging missions, our approach helps small swarms of robots to adapt more efficiently to changing environments, thus increasing their self-sustainability (survival rate). Finally, we propose a minimal version of ARTM, which is derived from the conclusions drawn through real-robot and simulation results.

Humans' adaptability to overcome the continuous changes of the environment is a result of their bodies' ability to change the way it interacts with the given situation. Therefore, augmenting robots with adaptable behavior, requires to exploit the possible dynamics emerging from the different interactions among all parts of their bodies. From this perspective, here in this pilot study, we developed a simple structure bipedal robot that enables a mutual interaction between its legs through an actuator network system (ANS). With this mechanism, the robot was enabled to transfer energy between its legs during locomotion. And as the experimental results showed, in terms of the roll motion and the vertical oscillation of the robot's body, the robot was able to produce a more stable locomotion behavior with mutually connected legs compared to other types of independent legs that do not have the ability to interact with each other.

For a natural social human-robot interaction, it is essential for a robot to learn the human-like social skills. However, learning such skills is notoriously hard due to the limited availability of direct instructions from people to teach a robot. In this paper, we propose an intrinsically motivated reinforcement learning framework in which an agent gets the intrinsic motivation-based rewards through the action-conditional predictive model. By using the proposed method, the robot learned the social skills from the human-robot interaction experiences gathered in the real uncontrolled environments. The results indicate that the robot not only acquired human-like social skills but also took more human-like decisions, on a test dataset, than a robot which received direct rewards for the task achievement., 34 pages

Recently, many humanoid robots have been developed and actively investigated all over the world in order to realize partner robots which coexist in the human environment. For such robots, high communication ability is essential in order to naturally interact with human. However, even with the state-of-the-art interaction technology, it is still difficult for human to interact with humanoids without conscious efforts. In this chapter, we apply the android robot, which has a quite similar appearance to a human, to a bystander in human-human communication. The android is not explicitly involved in the conversation, however makes small reactions to the behaviors of the humans, and the psychological effects on human subjects are investigated. Through the experiments, it is shown that mimicry behavior of the subject by the android is quite effective to harmonize the human-human communication.

The new era of cognitive computing brings forth the grand challenge of developing systems capable of processing massive amounts of noisy multisensory data. This type of intelligent computing poses a set of constraints, including real-time operation, low-power consumption and scalability, which require a radical departure from conventional system design. Brain-inspired architectures offer tremendous promise in this area. To this end, we developed TrueNorth, a 65 mW real-time neurosynaptic processor that implements a non-von Neumann, low-power, highly-parallel, scalable, and defect-tolerant architecture. With 4096 neurosynaptic cores, the TrueNorth chip contains 1 million digital neurons and 256 million synapses tightly interconnected by an event-driven routing infrastructure. The fully digital 5.4 billion transistor implementation leverages existing CMOS scaling trends, while ensuring one-to-one correspondence between hardware and software. With such aggressive design metrics and the TrueNorth architecture breaking path with prevailing architectures, it is clear that conventional computer-aided design (CAD) tools could not be used for the design. As a result, we developed a novel design methodology that includes mixed asynchronous–synchronous circuits and a complete tool flow for building an event-driven, low-power neurosynaptic chip. The TrueNorth chip is fully configurable in terms of connectivity and neural parameters to allow custom configurations for a wide range of cognitive and sensory perception applications. To reduce the system’s communication energy, we have adapted existing application-agnostic very large-scale integration CAD placement tools for mapping logical neural networks to the physical neurosynaptic core locations on the TrueNorth chips. With that, we have successfully demonstrated the use of TrueNorth-based systems in multiple applications, including visual object recognition, with higher performance and orders of magnitude lower power consumption than the same algorithms run on von Neumann architectures. The TrueNorth chip and its tool flow serve as building blocks for future cognitive systems, and give designers an opportunity to develop novel brain-inspired architectures and systems based on the knowledge obtained from this paper.

SANS has warned about the new variants of SSH dictionary attacks that are very stealthy in comparison with a simple attack. In this paper, we propose a new method to detect simple and stealthy attacks by combining two key innovations. First, on the basis of our assumptions, we employ two criteria: “the existence of a connection protocol” and “the inter-arrival time of an auth-packet and the next” . These criteria are not available, though, owing to the confidentiality and flexibility of the SSH protocol. Second, we resolve this problem by identifying “the transition point of each sub-protocol” through flow features and machine learning algorithms. We evaluate the effectiveness through experiments on real network traffic at the edges in campus networks. The experimental results show that our method provides high accuracy with acceptable computational complexity.

For a safe, natural and effective human-robot social interaction, it is essential to develop a system that allows a robot to demonstrate the perceivable responsive behaviors to complex human behaviors. We introduce the Multimodal Deep Attention Recurrent Q-Network using which the robot exhibits human-like social interaction skills after 14 days of interacting with people in an uncontrolled real world. Each and every day during the 14 days, the system gathered robot interaction experiences with people through a hit-and-trial method and then trained the MDARQN on these experiences using end-to-end reinforcement learning approach. The results of interaction based learning indicate that the robot has learned to respond to complex human behaviors in a perceivable and socially acceptable manner., 7 pages, 5 figures, accepted by IEEE-RAS ICRA'17

The complete quadratic combination (CQC) modal combination rule for seismic analysis of structures is a great achievement of Professor Armen Der Kiureghian. Among the other various uses of the CQC method, this chapter introduces the application of the CQC method to seismic response control of buildings with dampers to highlight the practical value of the CQC method. The expanded CQC method can accurately estimate the maximum response of buildings installed with response control damping devices and enables performance-based placement design of dampers. In this chapter, the performance-based design procedure of a viscoelastic damper (VED) is introduced for finding the storywise distribution of VEDs in a building such that each peak interstory drift coincides with the corresponding prescribed value. The mechanical properties of the employed VED’s dependence on amplitude and frequency of the excitation as well as material temperature are taken into account and a mechanical nonlinear four-element model that comprises two dashpot elements and two spring elements is proposed for the VED. The developed performance-based design procedure utilizes equivalent linearization of the VED and the expanded CQC method, which involves modal analysis with complex eigenvalue analysis. Seismic response analyses are carried out for high-rise building models with optimally placed VEDs, with the results demonstrating the effectiveness of the expanded CQC method and the validity of the proposed performance-based placement-design procedure.

For robots to coexist with humans in a social world like ours, it is crucial that they possess human-like social interaction skills. Programming a robot to possess such skills is a challenging task. In this paper, we propose a Multimodal Deep Q-Network (MDQN) to enable a robot to learn human-like interaction skills through a trial and error method. This paper aims to develop a robot that gathers data during its interaction with a human and learns human interaction behaviour from the high-dimensional sensory information using end-to-end reinforcement learning. This paper demonstrates that the robot was able to learn basic interaction skills successfully, after 14 days of interacting with people., Comment: The paper is published in IEEE-RAS International Conference on Humanoid Robots (Humanoids) 2016

Stochastic policy gradient methods have been applied to a variety of robot control tasks such as robot's acquisition of motor skills because they have an advantage in learning in high-dimensional and continuous feature spaces by combining some heuristics like motor primitives. However, when we apply one of them to a real-world task, it is difficult to represent the task well by designing the policy function and the feature space due to the lack of enough prior knowledge about the task. In this research, we propose a method to extract a preferred feature space autonomously to achieve a task using a stochastic policy gradient method for a sample-based policy. We apply our method to a control of linear dynamical system and the computer simulation result shows that a desirable controller is obtained and that the performance of the controller is improved by the feature selection.

Recently, robots are expected to support our daily lives in real environments. In such environments, however, there are a lot of obstacles and the motion of the robot is affected by them. In this research, we develop a musculoskeletal robotic arm and a system identification method for coping with external forces while learning the dynamics of complicated situations, based on Gaussian process regression (GPR). The musculoskeletal robot has the ability to cope with external forces by utilizing a bio-inspired mechanism. GPR is an easy-to-implement method, but can handle complicated prediction tasks. The experimental results show that the behavior of the robot while interacting with its surroundings can be predicted by our method.

A bio-inspired robot with many degrees of freedom (DOFs) might be beneficial in coping with various situations that occur in a real environment, because its physical structure resembles that of an animal it is modeled after. However, because of its complicated structure, it is difficult to explicitly model the dynamics and to design the control rules. In this study, we propose a predictive control method based on a non-parametric method. Instead of conducting parameter estimation for a certain parametric model, system identification is performed by collecting data. We apply our method to the control of a robot with a complicated structure. Experimental results show that the control of a robot with many DOFs can be achieved by the proposed method.

The compliance of muscles with external force and the structural stability given by biarticular muscles are important features of animals for realizing dynamic whole-body motion such as running and hopping in various environments. For this reason, we have been studying an electromagnetic linear actuator. This actuator emulates the behavior of a human muscle, such as spring-damper properties, through the quick control of output force, i.e., impedance control. It is expected to be used as an artificial muscle. In this paper, we design a monopedal robot possessing bi- and mono-articular muscles implemented by linear actuators. Thanks to the biarticular muscle, the direction of bouncing by a robot can be controlled by changing the stiffness ellipse at the endpoint, i.e., foot, of the robot. We make a simulator of the robot to evaluate dynamic characteristics and show that the robot hops stably by adjusting the stiffness ellipse. We also confirm that the behavior of the real robot is consistent with that of our simulator.

This paper presents a design strategy for simple robotic spines for androids with a natural upper-body postural appearance. The strategy allows us to develop a robotic spine with a minimum number of required movable joints, granting androids the ability to adapt natural upper-body postures while performing various gestures. In this study, by restricting the motion of the spine to sagittal plane for simplicity, the minimum number of required movable joints was determined using measured human spinal motion data. The error between the measured data and the model with a limited number of movable joints was calculated and the required number of joints was determined depending on its value. The efficacy of the proposed design strategy was validated through impression evaluation with five subjects. We also developed a prototype with four movable joints using low-cost mass produced parts.

The significant continuous increase in the power consumption of data centers, brought about by the adoption of cloud computing, is a major social problem. Moreover, the demand for data centers is expected to grow owing to the increasing requirement for data processing related to Internet of Things (IoT) applications. Consequently, reducing the power consumption of data centers has become an urgent and major issue. Modification of the operational settings of equipment residing in a data center leads to fluctuations in the power consumption of other equipment due to their complex interdependencies. Therefore, to reduce the total power consumption of the data center, it is not sufficient to reduce the power consumption of each piece of equipment, but one should consider the dependencies among the data center equipment. In other words, a coordinated control of the operational settings accounting for interdependencies among equipment is required to reduce the total power consumption of data centers. In this study, a data center energy management system focusing on the coordinated control of equipment in a data center is proposed. In the proposed system, the operational settings are determined using power consumption prediction that is based on the machine learning methods. We design a data center control system focusing on the task assignment of servers and air conditioner settings, and this system determines their settings based on a genetic algorithm. The proposed system is evaluated by experiments in our data center testbed and the evaluation results indicate that the proposed system reduces the total power consumption of the data center.

We propose a model for predicting the temperature distribution in data centers by using a convolutional neural network (CNN). Changes in the temperature distribution depend on the local structure of the data center, such as equipment locations and server types. Although the various physical relations in a data center were modeled as a network in our previous work, there were no mechanisms for automatically extracting the structure of the data center. The use of a CNN is a technique for learning local structure adaptively, which allows learning complicated features, such as the various physical relations in a data center. We evaluate the performance of the proposed model by using actual data from an experimental data center. The evaluation indicates that the proposed model can predict 20-minute future temperature distributions over 48 locations in 0.42 ms, with a root mean square error (RMSE) of 0.96 degrees. This accuracy is a dramatic improvement over simple linear prediction models, and the accuracy is sufficient to allow for control of air conditioners on the basis of these temperature predictions.

For the first time, by using a novel multiple individual bank sensing/writing and a memory bank interleave design, we demonstrate a double date rate 2 (DDR2) DRAM like interface phase-change memory (PCM) for M-type storage class memory applications . The write and read bandwidth is equal to 533MB/s, and the random read latency is 37.5ns, while the write latency is 11.25ns supporting a random write cycle of 176.7ns. In addition, a record high switching speed of 128ns with good resistance distribution is demonstrated with a super-fast Set material.

To reduce the power consumption in data centers, the coordinated control of the air conditioner and the serversis required. It takes tens of minutes for changes of operationalparameters of air conditioners including outlet air temperatureand volume to be reflected in the temperature distribution inthe whole data center. So, the proactive control of the airconditioners is required according to the prediction temperaturedistribution corresponding to the load on the servers. In thispaper, the temperature distribution and the power efficiencyof air conditioner were predicted by using a machine-learningtechnique, and also we propose a method to follow-up proactivecontrol of the air conditioner under the predicted optimumcondition. Consequently, by the follow-up proactive control ofthe air conditioner and the load of servers, power consumptionreduction of 30% at maximum was demonstrated.

This paper deals with a multiagent path-planning problem where several robots track humans to obtain detailed information on human behaviors and characteristics. For this, agents' paths are planned on the basis of the similarity between the predicted positions of humans and the agents' field of view. The long-horizon path planned on the basis of an accurate long-horizon prediction improves the tracking performance. However, it requires heavy computation and is less useful if the prediction is inaccurate. Since the accuracy of the prediction depends on the situation, the prediction term is determined by the similarity between the current and previous predictions. The results of computer simulation showed that our path-planning method works well for trajectories of humans in a dynamic environment by changing the horizon length of the path planning.

Interpretation of the living organism’s ability to utilize noise from engineering point of view may be beneficial for realizing a simple, yet adaptive, robotic system. This paper presents a bacteria-inspired mobile robot, with a 1-DOF motor and a single sensor. Even with such limitations, the underactuated mobile robot is able to navigate toward a gradient-inducing goal in a two-dimensional space by properly utilizing environmental noise that directly affects its motion. The way the robot utilizes this external noise to control its navigation behavior is interpreted based on biological fluctuation: a simple perspective commonly used to describe how living beings utilize internally generated biological noises.