ABSTRACT. Ants live in organized societies with a marked division of labor among workers, but little is known about how this division of labor is generated. In the first part of this talk I will present network analyses based on more than 9 million interactions to show how workers move from one behavioural group to another as they age. I will also present a model allowing one to predict the pattern of behavioural maturation of workers based on who they interact with. In the second part of the talk I will outline the conditions necessary for cooperation and communication to evolve among evolving groups of robots and report a few surprising results.

ABSTRACT. We design a new type of a two-link underwater snake robot using screw-drive units which are connected by an active joint. The screw-drive mechanic unit provides the snake robot with the capability of omnidirectional locomotion. In this paper, we focus on the development of the robot and realize the underwater locomotion in a 2D plane by teleoperation. Basic experiments for different types of locomotion are carried out.

ABSTRACT. Maintenance of huge complex structures such as chemical plants, bridges, and towers is an expected task for robots. In these structures, robots have to climb various columnar objects in order to conduct their work. In this research, we propose a four-legged robot that climbs unknown columnar objects by applying a mechanism inspired by the octopus. We developed a prototype robot, and experiments were carried out. As a result, we confirmed that the developed robot could climb both an unknown column pipe and a square pillar.

ABSTRACT. Collective decision making is a consensus building mechanism inspired by the collective behaviour of social insects. By considering the collaboration of this mechanism with the supervised learning technique, this study focuses on a collective learning mechanism using multiple Deep Neural Networks. The collective learning methodology is also becoming one of the hot topics recently, in which many studies basically focus on collaborative unsupervised learning. In this paper, the authors show the fundamental idea of collective semi-supervised learning with Deep Neural Networks based on the collective decision making mechanism. Future contributions are discussed from two different points of view, the scientific interest and the engineering application.

ABSTRACT. The automated driving is an emerging technology in which a car performs recognition, decision making, and control. The decision-making system consists of route planning and trajectory planning. The route planning optimizes the shortest path to the destination like an automotive navigation system. According to static and dynamic obstacles around the vehicle, the trajectory planning generates lateral and longitudinal profiles for vehicle maneuver to drive the given path. This study is focused on the trajectory planning for vehicle maneuver in urban traffic scenes. Numerical simulation shows that the proposed planner controls the vehicle properly according to surroundings.

ABSTRACT. In this study, we developed a circular arc gear-type modular robot equipped with an inter-module communication function. While a wide range of modular robots has been developed, this is the first of its type to allow inter-module communication. We experimentally confirmed that communication between multiple modular robots could be achieved using the wireless communication function of a ZigBee device.

ABSTRACT. This paper proposes odometry for a snake-like robot equipped with tactile sensors. Teleoperated snake-like robot is often invisible in a narrow space. In order to visualize the robot on the remote display in the third person view, we propose odometry using only local sensory information. The key point of the method is the use of tactile sensors that locate the contact point. Under the assumption that the contact points does not move in sidwinding gate or in crawler gait, the head point in the world coordinate system can be calculated from the position of contact points and joint angle of the robot. The proposed method is verified with a simulator with physics engine.

ABSTRACT. The African dung beetle Scarabaeus galenus can use its front legs to walk and manipulate or form a dung ball. The interesting multifunctional legs have not been fully investigated or even used as inspiration for robot leg design. Thus, in this paper, we present the development of real dung beetle-like front legs based on biological investigation. As a result, each leg consists of three main segments which were built using 3D printing. The segments were combined with in total four active DOFs in order to mimic locomotion and object manipulation of the beetle. Kinematics analysis of the leg was also performed to identify its workspace as well as to design its trajectory. To this end, the study contributes not only novel multifunctional robotic legs but also the methodology of the bio-inspired leg design.

ABSTRACT. In the last few decades, much effort has been devoted to the research of robots that can complete dangerous tasks on behalf of a human. Among these robots, snake robots have been developed actively because of their potential flexibility and robustness. However, there has been little research on how to handle a fault in a snake robot. To continue the task under failure, we must first know the detail of the failure, and then, based on this knowledge, remake the model of the robot. This paper, as the first step, proposes a method to diagnose the fault status using the current data of a test motion, when malfunctioning of an actuator is detected. As a typical situation for a snake robot, we consider when the robot is rolling itself around a pipe. In spite of rather limited movement of the robot in the situation, experimental results show that our method can diagnose the failure status. Moreover, the results suggests a way to improve our method.

ABSTRACT. A snake robot that we have developed can move the inside and outside of a pipe by using helical rolling motion. This capability is expected to be useful for pipe inspection. In research on snake robots, it has been studied to adapt reaction force from environment. However, adaptation of reaction force from environment during helical rolling motion has not been realized. Therefore, in this paper, we developed a snake robot that can measure pressure force on the entire circumference of the snake robot's body by using Center of Pressure sensor. In addition, we proposed a method to recognize a bending pipe during helical rolling motion. In the method, the snake robot recognizes arrival to a bending pipe and bending direction of a bending pipe by using data of Center of Pressure sensor and IMU. The proposed method is verified by experiments using the developed snake robot.

ABSTRACT. Bio-inspired robotics is an approach that looks at how nature solves a (complex) problem and applies the solution to a robot. Based on this approach, we have looked at walking animals that can exploit their legs for multiple functions, like locomotion and object manipulation/transportation. This is to expand the ability of walking robots. According to this, we have developed our robotic model based on a dung beetle, an animal that can walk and roll a dung ball using its legs. In this paper, we present distributed sensor-driven control for generating dung beetle-like walking and ball rolling behaviors of the developed model. The control mechanism is based on the Walknet bio-inspired controller. Furthermore, this paper gives a proof of concept that locomotion and object manipulation/transportation can be achieved without installing additional manipulators and/or grippers on a robot as shown by other works. To this end, the bio-inspired strategy can avoid the requirement of additional energy to power the manipulator or gripper system; thereby leading to an energy-efficient multi-functional robotic system.

ABSTRACT. This paper investigates the modeling and control problem for a planar two-link rigid-flexible manipulator. A partial differential equation (PDE) model which only contains measurable states in the boundary conditions is derived by the Hamilton’s principle. Based on this model, a feedback boundary control scheme is proposed to regulate joint states reaching desire and realize vibration rejection on the flexible link. The asymptotic stability of the closed-loop system is ensured by the Lyapunov method theoretically, and the numerical simulations verify the performance of the controller as well.

ABSTRACT. Hyper-realistic Humanoid Bio-robotic Systems (HHBS) are the precise electro-mechanical bodily emulation of natural human being in materiality, form and function. However, the standardised approach of constructing static ocular prosthesis for implementation in modern HHBS design, contradicts innate organic human criterion as the artificial eyes are void of the intricate dynamic fluidic functions of the natural human iris. The aim of this paper is to outline the development and construction process of a pair of realistic artificial humanistic optical retinal sensors that accurately simulate the autonomous fluctuating operations of the human iris in reaction to visceral emotion and photo-luminescent stimuli and retain the optical sensory capability and integral aesthetic materialism of the organic eye. The objective of the auto-dynamic pupillary framework is to advance the external expressive / embodied realism of HHBS towards achieving a more accurate operational and embodied simulation. Prospective future application and advancement of the outlined optical system presents potential implementation in the field of medical ocular prosthetic design, with the aim of enhancing the naturalistic operations of future fabricated human eye replicas, thus conceivably reducing the malaise and discomfiture commonly associated with the archetypical fixed artificial eyes.

ABSTRACT. A typical approach when designing a bio-inspired robot is to simplify an animal model and to enhance the functionality of interest. For hexapod robots, this often leads to models with an unknown performance; thereby influencing their functionality and requiring supplementary mechanics to become multifunctional. A preferable solution is to employ the embodied multifunctional capabilities of the animal as inspiration for robot design. For this reason, we present a method for translating the kinematic chain of a dung beetle, from which an accurate model and a simplified one were created and simulated. A comparison between the two modeling approaches shows a similar performance in regards to walking stability and accuracy, but differences when it comes to speed and multifunctionality. Here the accurate model outperforms the simplified model, by both walking faster and being capable of performing additional locomotory tasks. In conclusion, the accurate model of a dung beetle-inspired robot is advantageous in regards to multifunctional abilities including walking as well as standing on and rolling a ball, like a dung beetle.

ABSTRACT. Most studies on self-organized flocking of robotic swarms are based on a computer animation model named Boid. This model reproduces flocking motion by three simple behavioral rules: collision avoidance, velocity matching, and flock centering. However, the flocking performance largely depends on how these rules are configured, and there is no guideline for the configuration. This paper introduces hierarchical interaction based flocking by employing individuals that can switch their roles. Robots can move as (sub-)leaders or (sub-)followers depending on the situations. The flocking performance is evaluated, and swarming behavior is analyzed in a scenario where robots consecutively travel between two landmarks.

ABSTRACT. In recent years, some studies have been conducted on algorithms to construct various shapes autonomously using a large number of robots. In a related study by Rubenstein , they use 1024 KILOBOTs and study algorithms that can self-assemble into various forms without centralized control. Because such research uses many robots, an external power source by power line is not realistic, therefore it is inevitably driven by batteries. However, with battery driving, operation time is restricted depending on the size of the battery, making it difficult for long-term experiments. In response to this problem, we apply a newly developed two-dimensional communication system to KILOBOT, thereby eliminating the need for a battery and propose a new wireless power supply system for robots that excludes restrictions on operating time. In this research, we focused on an ultra-small robot called KILOBOT, the trial production of an ultra-compact wireless power supply system, which makes the operating time practically infinite, and we performed rudimentary operation testing. As a result, it was determined that a sufficient power supply to operate a single KILOBOT can be achieved.

ABSTRACT. Value-sensitive decision strategies, inspired by honeybee nest site selection, are a promising approach to robotic swarm collective decision making. Recent research has revealed these strategies to be influenced by site location, which biases decisions in favor of closer sites. This bias is advantageous when site evaluations incorporate distance into site quality, but hinders accuracy when site evaluations are required to ignore distance. Two extensions of an existing decision making strategy are presented in order to efficiently reduce distance bias in robotic swarm site selection: interaction delay and interaction frequency modulation. A single model combining these techniques improves decision accuracy by 15% overall and by 60% when the better site is at a further distance. Compromises in decision time, predictability, and flexibility associated with each proposed extension are analyzed to guide designers in selecting the best mechanisms to improve decision accuracy in swarm site selection scenarios.

ABSTRACT. We present a model investigating the effect of agent-to-agent energy transfer or ”feeding” on the behaviour of a group of robots that source their energy from their surrounding environment. We show increased mobility of the swarm as a result of trophallaxis in long term simulations, where the period of a single trophallaxis event is shorter than the time required to convert energy from the surrounding environment. In contrast to previous work, we investigate the case of a finite source from which the robots gain energy. The results suggest that the centralised configuration of both ”food” and robots is a limiting factor in swarm performance in terms of the area of the model space explored.

ABSTRACT. Collective animals, such as flocking birds and schooling fish, organize their synchronized behavior as a single ordered group without a particular group leader. In the basic models of collective behavior, three types of rules are considered as fundamental ingredients: the alignment rule by which an agent matches its velocity with those of others in its neighborhood, the attraction rule by which an agent approaches its neighbors if they separate in its neighborhood, and the repulsion rule by which an agent separates from its neighbors if they are too close to each other in its neighborhood. In these models, agents assemble as a single group based on the attraction rule, with directed motion based on the alignment rule, and maintain stable inter-individual distances based on the repulsion rule. Then, the geometrics of the group achieve a stable lattice-like interaction network. Recent studies, however, show that although animals in a collective group can exhibit global order, there are diverse individual movements within the group, which would derive from ceaseless negotiations with respect to attraction-repulsion decision making among individuals. In this study, we used soldier crabs (Mictyris guinotae) who live in tidal flat and form huge and dense swarms around the ebb tide. We tracked individual trajectories in laboratory condition and investigated how they balance between attraction and repulsion actions within the group. Here we defined that the individual choice is an attraction action if the individual moves to the major area, that is, the side of the neighborhood where the number of neighbors is larger than that on the opposite side (i.e., the minor area), and a repulsion action if the individual moves to the minor area. First, we show that individuals’ attraction-repulsion actions display various behaviors depending on different configuration of neighbors; they seem always approach to major area in their neighborhood, but rather, avoid the area if it is crowded excessively. This suggest that an individual can anticipate avoiding a crowded situation. Moreover, when we defined the duration by examining whether an individual switched its action from attraction to repulsion or vice versa between time steps, we found that the duration obeys a power-law distribution. These results suggest that a flexible and robust swarm can be achieved through the critical behavior of individuals’ attraction-repulsion decision making within the group.

ABSTRACT. Collective animal behavior has attracted much attention in various fields from physics to behavioral ecology, robotics and social science. Previous studies on collective behavior have focused on external factors including interactions among individuals and responses to the environment, and have revealed how collective behavior emerges from interactions between individuals and evolutionary reasons why animals exhibit collective behavior in terms of evolution. In the context of ecology of the solitary animals, it has been shown that animals often exhibit patterns of spontaneous activity patterns that can lead to efficient movement patterns even in no-stimulus conditions, suggesting that individual spontaneous behavior can be crucial for collective dynamics. However, spontaneous behavior in social animals is poorly understood. Here, we analyzed the time series data of ant individuals’ movements and identified the characterized their spontaneous behavior. We found that ant individuals exhibit behavior patterns that fit the power law that and there are caste differences of burstiness and memory parameters between castes. Even in social animals, the spontaneous individual behavior corresponds to universal power law patterns of behavior and the characteristics of these patterns depend on behavioral castes, which could lead to efficient information transfer and division of labor.

ABSTRACT. In social insects, workers within a colony perform different tasks that contribution to the functioning of a highly organized system as a colony. To adapt to different task requirements, communications is important for workers. In this study, we focused on a basic element of behavior, circadian rhythm, and investigated the effects of interaction between adults on activity rhythms using the monomorphic ant species Diacamma sp. To this end, we devised an automatically tracking system that detects color, which enabled us to collect long-term series data on the movement of individual ants within groups. Behavioral tracking revealed that solitary young workers (YWs, nurse age workers) and old workers (OWs, forager age workers) showed circadian rhythms under the constant light condition. Under in-group conditions, adult-adult interaction affected circadian activity rhythms but it was highly dependent on the ages of interacting ants. This result suggests that different ages of ants may give different kinds of information. We speculate that this age-dependent response variability may have relevance to their task performances. In future studies, it will be desirable to investigate the relationship between activity rhythms and task performances under more natural condition such as under field or colony conditions.

ABSTRACT. Dictyostelium discoideum is a social amoeba exhibiting distinct self-organising behaviour at different phases of its life - moving in a coordinated way towards areas with food, signalling lack of food and recruiting partners to create a single super-organism (fruiting body). These phases and their underlying mechanisms are excellent models useful for understanding other natural cells’ behaviour (e.g. cancer cells), as well as to engineer artificial systems such as swarms of robots. This paper focuses specifically on the aggregation phase of Dictyostelium discoideum. We present a detailed agent-based - “bottom-up” - model, which exhibits a series of individual, collective behaviours and emergent properties of social amoeba Dictyostelium discoideum. We extended previous models of the aggregation phase with: a pre-aggregation phase; and three different levels of quorum sensing allowing collective decisions to be taken in a decentralised manner for: (1) identifying the time for aggregation phase; (2) providing aggregation territories of homogeneous size; (3) allowing the appearance of late centres. The key character of our model is the cells’ self-assessment and self-generated gradients arising from six chemical factors: PSF, CMF, Adenosine, cAMP, PDE and CF released by each individual amoeba. We programmed our model in Matlab. Our results show a series of behaviour close to individual and collective behaviour of living Dictyostelium discoideum. In particular we observed: inhibition of centers too close to each other; appearance of late centers when aggregation territories are too large; and measured homogeneous size aggregation territories emerging from the cells’ behaviour. Future work will address the remaining phases of Dictyostelium discoideum life cycle (migration and culmination).

ABSTRACT. We conducted quantitative analysis of flock of geese flying in V-shaped formation to understand the regulatory mechanism of positioning in the flock and timing of wingbeat for upwash exploitation by focusing on the phasing relationship of individual wing flap in the viewpoint of a couple oscillator system.

ABSTRACT. The ant system is well known as one of self-organized systems. Complex collective behaviours are emerged even though individuals are interact with each other using simple rules. Droplets of ants are observed when they construct chains or bridges. Several studies reported that droplet dynamics of ants resembled chaotic behaviours of dripping faucet. However, little is known how individual decision-making determine collective droplet dynamics. In this paper, I developed an agent-based model in which each artificial ant tuned their behaviours based on local environments, resulting in complex droplet dynamics. I found that my model partly represented typical properties of droplet dynamics of real ants.

ABSTRACT. 1. INTRODUCTION An important question of swarm intelligence is the connection between individual-level- and group level-performance. We focused on the colony-level foraging activity of the ponerine ant, Diacamma sp. from Japan, in which individual-level biological information, such as on circadian rhythm and reproductive physiology related to foraging, has been recently extensively gathered [1, 2, 3]. In ants, foraging activity is known to be influenced by various factors, such as time of the day, temperature, relative humidity, light intensity, and food availability [4, 5, 6]. We tested the following four patterns: (1) the circadian rhythm, known in isolated individuals in the laboratory, strongly influences colony-level foraging [2, 3]; (2) foraging activity has seasonality, (3) colonies send more foragers to the outside when foraging efficiency is temporarily high, or alternatively colonies send more forager when the foraging efficiency is decreasing simply due to hunger; and (4) environmental factors such as low temperature limit foraging activity.

2. METHODS All field works have been performed at Sueyoshi Park in Naha city Okinawa, Japan. To assess daily foraging activity, we monitored a total of 13 colonies (spring: 3); (summer: 4); (autumn: 5); and (winter: 2). We observed all incidences of workers leaving and returning to the nest (bringing with or without things) for 24-hours. We defined the hourly foraging efficiency as (the number of returning foragers carrying food × 100)/the total number of returning foragers in each hour. We defined the hourly foraging activity as the number of leaving plus that of returning foragers in each hour. Air temperature (°C) was also recorded by using a digital thermo hygrometer.

3. RESULTS Foraging of D. sp. colonies were active around-the-clock except in winter. However, they were predominantly diurnal with more frequent foraging in the day time than at night regardless of the seasons. Temperature itself did not directly influence foraging activity, but low temperature at night seemed to lead to lower foraging activity and efficiency. Foraging activity was positively related to foraging efficiency with a large seasonal variation.

4. DISCUSSION The colony-level foraging activity tended to be diurnal by sending more foragers during the day time. This finding was somewhat different from the previous study of this species in which foragers were active around-the-clock in the laboratory [2]. It might be due to environmental factors such as temperature and humidity that may affect the rhythms of foraging activity. In this study, we also found interaction effect of temperature on time of foraging activity. Foraging was totally stopped during the night in winter but was improved in spring, summer and autumn. Foraging or the other outside-nest activities were facilitated by the higher temperature during the night. Colonies sent more foragers when foraging efficiency was high; however, this relationship was weak and varied seasonally. All above patterns vary to some extent in non-ponerine ants which has distinct morphological castes, suggesting that it will be a good opportunity to understand relationship between foraging system and social system in Ponerine ants.

5. CONCLUSION We highlight the finding that can contribute future studies on the mechanisms of collective decision making.

REFERENCES [1] K. Tsuji, N. Kikuta & T. Kikuchi, “Determination of the cost of worker reproduction via diminished lifespan in the ant Diacamma sp.”, Evolution, 66, 1322–3311, 2012. [2] T. Fuchikawa, Y. Okada, T. Miyatake & K. Tsuji, “Social dominance modifies behavioral rhythm in a queenless ant’, Behavioral Ecology and Sociobiology, 68, 1843–1850, 2014. [3] H. Fujioka, M. S. Abe, T. Fuchikawa, K. Tsuji, M. Shimada & Y. Okada, “Ant circadian activity associated with brood care type”, Biology Letters, 13, 20160743, 2017 [4] J. Medeiros, A. Araújo, H. P. F. Araújo, J. P. C. Queiroz & A. Vasconcellos, “Seasonal activity of Dinoponera quadriceps (Formicidae: Ponerinae) in the semi-arid caatinga of NE Brazil”, Revista Brasileira de Entomologia, 56, 81–85, 2012. [5] L. Ríos-Casanova, G. Castaño, V. Farías-González, P. Dávila & H. Godínez-Alvarez, “Activity patterns of the red harvester ant in a Mexican tropical desert”, Sociobiology, 61(2), 133–135, 2014. [6] W. G. Whitford, W. G., “Seasonal and diurnal activity patterns in ant communities in a vegetation transition region of southeastern New Mexico (Hymenoptera: Formicidae)’, Sociobiology, 34, 477–491, 1999.

ABSTRACT. In this paper, we propose an extremely simple decentralized control scheme for swarm robots that can perform spatially distributed tasks in parallel. Each agent has an internal state called “confidence.” Each agent first moves randomly to find a task. When it finds a task, its confidence increases, and then it attracts its neighboring agents to ask for their help. Simulation was used to demonstrate the validity of the proposed control scheme.

ABSTRACT. We introduce an experimental apparatus to make an adaptive behavior ineffective and report results of preliminary experiment. Soldier crabs wander en masse on intertidal flat surface during the daytime low tide. They can burrow into the ground when a person approaching them. Though their behavior seems to be adaptive, the process is unclear. To understand it, we considered that it is necessary to investigate how they cope with situation where digging behavior unfortunately does not result in hiding. Digging is usually accompanied by hiding. We break this causality by developing thin soil layers. In the assembled apparatus, crabs could dig the soil layer and then go down to the next layer. The behaviors in case of one or two crabs implied some effect of presence of neighbor.

ABSTRACT. In this paper, we address stability analysis for swarm robots with heterogeneous abilities such as velocity and sensing region. We previously proposed a decentralized navigation method for heterogeneous swarm robots. With the method, connectivity maintenance is achieved while satisfying physical constraints which are unique to each robot. We gave a parameter in the control inputs of follower robots to change swarm shape. Here we focus on the shape of the whole swarm and mathematically prove that the shape eventually converges to equilibrium state. Furthermore, we show simulation results to confirm the convergence.

ABSTRACT. First, generalization of the funnel control based coupling law is provided to achieve a practical synchronization. In addition to this, we detail the discussion on the synchronized behavior of heterogeneous agents under funnel coupling. In particular, emergent dynamics has been conjectured, which can predict the synchronized behavior by its solution trajectory. Discussion and comparison associated with the similar concept named blended dynamics in strong coupling technique have been given, and also the effectiveness of our conjecture is illustrated by simulations.

ABSTRACT. This paper addresses a formation control problem of multi-agent systems allowing a scale freedom. By considering the scale freedom in addition to translation and rotation ones, agents can achieve a desired formation more flexibly. A distributed control input to solve this problem is designed based on the gradient-flow approach. Its effectiveness is illustrated through some simulation results.

ABSTRACT. We present here collective dynamical orderings observed in adroplets system where oil droplet surfers floating on the surface of aqueous surfactant solution. The droplets are self-propelled by the so-called solutal Marangoni effect and forms clusters periodically changing their shape. This cluster dynamics was overved only when Marangoni effect mostly decayed. Experimental evidences suggest that the interaction among droplets is essential for the cluster dynamics. We propose that our simple droplets system can be a model for complex swarming behaviors of living organisms.

ABSTRACT. The behaviour of groups of decanol droplets placed in a thin layer of aqueous sodium decanoate solution was investigated experimentally. When the droplets were placed in decanoate solution, they self-propelled on the surface and first repelled then to attract each other. The experiments show a variety of collective phenomena in the different time scales (in the order of a few seconds to few hours). The phenomena are dependent on initial conditions, such as the thickness of the solution and the size of the system. We report this complex collective dynamics and discuss the underlying potential mechanism in the talk.

ABSTRACT. Cultured neural systems are much simpler than real brain systems, however, they have essential properties, including spontaneous activity, various types and distribution of cells, high connectivity, and rich and complex controllability. Previous works show that dissociated cultured neurons can learn a simple task as if it behaves to avoid a stimulation from outside. By simulating a simple model neural network, we show that spiking neural networks with spike- timing dependent plasticity (STDP) can reproduce such learning results. Recently, we extended the phenomenon to larger networks. We name this principle ”learning by stimulation avoidance” (LSA). The efficiency of LSA is proved by computer simulation, and the next step is to see whether the same mechanism actually works in a biological neural network or not. In this study, we conducted learning experiments using smaller-sized cultured neuronal cells than previous works, and this revealed that such learning results can be scaled to small network in vitro like the results in silico. In addition, the results of neural dynamics analysis suggest that cultured neurons change the network structure with synaptic plasticity to achieve a learning task rather than merely increase whole firing rates. It also suggest that LSA based on a synaptic plasticity can actually work in biological neural networks.

ABSTRACT. In this paper, we develop a reconstruction method based on rank minimization. This is based on the AR model on a graph signal and achieves high quality reconstruction. First, we define an AR model and structured matrix for graph signals. Next, we formulate a restoration problem and propose a rank minimization approach. Finally, the proposed method is demonstrated.

ABSTRACT. This paper is concerned with distributed cooperative control of linear uncertain agents with constraints on the amplitudes of inputs and outputs. We derive a condition for achieving both state synchronization and input/output constraints in the presence of norm-bounded uncertainties in the agent models. We develop a synthesis method of the desired state feedback gain in terms of linear matrix inequalities based on the notion of positively invariant sets. The desired feedback gain can be efficiently obtained by solving a certain convex programming whose size is independent of the number of agents.

ABSTRACT. This paper deals with MPC-based robust and optimal control problem of the traffic flow on highways with uncertainty by computing the variable price in each road section. At first, the model of the transportation network assuming application to a complex road network and the driver behavior model for variable pricing are discussed. Then, in order to optimize highway traffic flow, the variable pricing control problem is formulated based on the model predictive control framework that takes price variation and prediction error into account for the obtvariable price setting method is more robust than the conventional methods for the highway network system with uncertainty.

ABSTRACT. This paper proposes novel distributed proximal gradient algorithms, called Consensus-based Iterative Soft-Thresholding (C-IST) and Consensus-based Iterative Hard-Thresholding (C-IHT),for sparse optimization over connected networks. At each iteration of C-IST and C-IHT, agents run the consensus algorithm based on local communications with neighbors and evaluate a proximal operator to minimize the own objective function. We show by simulation that C-IST and C-IHT can find an approximate solution of sparse optimization problems in a distributed way.

ABSTRACT. Simulation of a million flocks is studied by using a simple boid model. It is modeled with a differential equation on the 3D space. Flocking pattern in general is a function of both space and time scales, and it is not a mere aggregation process in a physical space. We also calculate the fluctuation of speed and the local number for each flock size and report the characteristics of fluctuation dependent on flock size. This paper can clarify the hierarchical nature of flocking.

ABSTRACT. In social tagging systems, service users create a number of tags, each of which is reused according to people's preferences. As a result, the diversity of tag vocabulary continues to increase as they are exposed to selection pressure from cultural preferences of the time. This is analogous to living ecosystems in nature. We address the question how new words arise and get accepted by crowds in terms of community structure of the network defined by the vocabulary usage of users. The result suggests that a new word, or novelty, that would be accepted by other users tend to come from the peripheral part of the network.

ABSTRACT. When we separate an ant group into a lazy ants group and a diligent ants group, some lazy ants become to be diligent in a lazy ants group, some diligent ants become to be lazy in a diligent group. This behavioral change can be well explained by Response Threshold Model. However, there is not enough argument whether the Response Threshold Model can be applied when the separated ants groups are rcombined to one group again. We observed behavioral change of each individual ant when recombining the separated ant groups in Camponotus Japonicus. We confirmed the behavioral changes that can be explained by response threshold model in the lazy ants group.

ABSTRACT. We observed several queenright colonies of the ant, Campnotus janpnicus, using RFID sensors, which can record time when an ant passes under the sensor and unique ID of each ant. Using recorded data, we analyzed the activity distributions and the correlation between the data collected at different positions in the multiple nest cases. The analysis represents that the difference of location where the ants perform their tasks in the nests affects the statistical aspects of the ant behavior, and it could be concluded that task allocation relates to spatial position in long-term observation.