ABSTRACT. The last few years have seen significant progress in our understanding of how one should structure multi-robot systems. New control, coordination, and communication strategies have emerged and, in this talk, we discuss some of these developments. Inspiration will be drawn from naturally occurring, self-organizing systems and we will show how one can go, in a provably correct manner, from global, team-level specifications to local coordination rules for achieving and maintaining formations, area coverage, and biologically inspired swarming behaviors.

ABSTRACT. Groups of individuals in social insects form cooperative societies or "superorganisms". However, it is also known that evolutionary conflicts which can lead to "the tragedy of the commons" are widespread in those societies. A major question in evolutionary biology is how resolution of conflicts among the lower-level units was achieved to form the higher unit. Policing is an important mechanism that enables the formation of group harmony by preventing selfish behavior of the lower units. Ohtsuki and Tsuji (2009) has proposed a new hypothesis on the mechanism of evolution of worker policing in social Hymenoptera. A novel prediction of this model is that worker policing will depend on the stage of colony growth and under queen single mating strong worker policing will occur only at the ergonomic stage. In this symposium I will talk about two topics related to this hypothesis. First, I show data strongly supporting Ohtsuki and Tsuji's prediction in the ponerine ant Diacamma sp. (Tsuji et al. unpublished data, Tsuji et al. 2012). Second, I suggest a proximate mechanism of colony size related behavioral regulation in this ant that is achieved though individual contacts. This is a pioneering study revealing an explicit colony size "sensing" mechanism in social insects.

ABSTRACT. We investigate the optimal sampling times of school-of-fish social behaviours for a multi-agent source localization problem. We explicitly include the neighbour alignment social behaviour and investigate its relative role in terms of the arrival-time performance. For two different individualistic behaviours, results show that the social behaviours work significantly better when their operation is coupled together at the same time-scale and decoupled from that of the individualistic behaviour. The proposed coupling of the social behaviours results in a significantly better mean arrival-time performance and also in reduction of arrival-time uncertainty.

ABSTRACT. In this paper, we propose a multiple robot control system working in unfamiliar or unstructured environment. We focus on a robust model predictive control (robust-MPC) that enables robots to perform in uncertain environments and demonstrate that the performance of the robust-MPC is dependent on the amount of uncertainty that derives from sensor noise. The proposed system includes cooperative observation based on the information seeking theory. Therefore, the system works for not only treating the uncertainty but also decreasing it. A simulation result of a case study of collision avoidance shows that the combination of robust-MPC and cooperative observation enables the robot to move efficiently and arrive at the goal faster than the single robot case.

ABSTRACT. This paper focuses on an improved response threshold model for effective task allocation in multi-agent system. Each agent has response thresholds for tasks and regulates its thresholds adaptively according to not only its performance but also social interaction based on desynchronization method using pulse coupled oscillator models. The proposed algorithm is implemented in simulated robots and experimental results show the improved performance than the original response threshold model.

ABSTRACT. This paper suggests a dispersion robot system for surrounding the domain where targets crowded. In related works, it was able siege only multiple targets individually. Here, we propose a new method which surrounds collectively multiple targets.

ABSTRACT. In this paper, we consider the problem of alignment forming in multi-agent systems based on linear regression of the agents’ positions. The motivation comes from a multi-agent source seeking problem requiring the knowledge of the direction of movement shared by all agents. We explore two different methods to address this problem. For the first one, we employ average consensus to perform linear regression locally over the network while in the second, we employ a distributed optimization technique in local least square estimates performed by each agent using only relative positions of its neighbors. Through numerical examples, we verify the effectiveness of both methods in the context of source seeking.

ABSTRACT. Red harvester ants allocate themselves some tasks. In this paper, we proposed a model which represents their macro behaviors. In this model, colony size fluctuates due to workers' birth and death. Using a simulator based on this model, we investigate possibility that red harvester ants change their behaviors based on global information, which is colony size and amount of food of colony.

ABSTRACT. Self-assembly of homogeneous components has the advantage of being a decentralised and highly parallel method for assembling multiple target structures, and is ideal for effective large-scale manufacturing. Yet assembly yield may be negatively affected by the formation of incompatible substructures that prevent the formation of complete target structures. In this work we present physical and theoretical analysis of a simple magnetomechanical self-assembling systems exhibiting the problem of incompatible substructures in the formation of closed circular target structures out of eight homogeneous components. The assembly yield of physical experiments from 8 to 40 components is compared with the predictions of a computational model, and the model is found to accurately predict both the mean and standard deviation of the experimental yield.

ABSTRACT. This paper discusses a novel class of Particle Swarm Optimization called Attractive-Repulsive Group-based Particle Swarm Optimization (ARGBPSO). This algorithm is used for performing velocity profiles on the three-degree Bezier-based paths. Constraints of maximum allowable radial and tangential accelerations and tangential velocities are considered. The proposed method is designed for performing minimum-time collision-free trajectories in a multiple-vehicle system. The problem of minimizing the reaching time of the slowest vehicle is addressed. A particle group represents a set of particles containing path’s two control points of each vehicle. The term of attractive and repulsive forces on the particle groups are introduced. The searching process executed by the GBPSO can be described as searching the suitable control points that perform minimum time trajectories. The first and last two control points are used as the state vector of a single particle. The proposed method has advantages in shortening velocity profile generation time and thus enhances the searching time. The results of a simulation demonstrating the performance of the proposed GBPSO also are presented.

ABSTRACT. Many biological and artifact networks represent modular structures, in which the network can be decomposed into several subnetworks. How modular networks are evolved is still open discussion. In this presentation we study the simple model by evolutionary simulations for the modular network evolution based on robustness to noises, particularly, non-linearity noise reduction in a node-activity. The model suggest that it is possible to evolve modular networks under conditions; if the given goals to networks, input and target output pairs, involve modular feature; and if the signal transfer in a node is carried out with non-linear manners in the sense of a saturation at upper and lower bounds. Our model highlights the positive role played by noises in modular network evolution.

ABSTRACT. In this paper, we provide a new approach to classify and recognize the acoustic events for multiple autonomous robots systems based on the deep learning mechanisms. For disaster response robotic systems, recognizing certain acoustic events in the noisy environment is very effective to perform a given operation. As a new approach, trained deep learning networks which are constructed by RBMs, classify the acoustic events from input waveform signals. From the experimental results, usefulness of our approach is discussed and verified.

ABSTRACT. In this research, we assume environments that occurrence of a huge disaster that cannot utilize communications infrastructure, and then, to cope with such problem, we focus on the SpotPass communication by refuge's mobile terminal. We verify a sharing possibility of the global crowd movement information by exchanging information of each terminal keeping through simulations. As a result, it became clear that existence of refuges who moves to shelters except a nearest shelter achieves the bridging function of the information and has a big influence on the information sharing rate.

ABSTRACT. We constructed an agent-based simulation using GPU with an asynchronous evolution algorithm. The simulator can handle predator-prey relationships. We describe the simulator, and introduce some snapshots of the results.

ABSTRACT. Swarm robotics (SR) is a novel approach to the coordination of large numbers of homogeneous robots. After generated collective behaviors through local interactions between robots and their environment, we are now focused on how task allocation occurs during SRS's operation. In this paper, a clustering method inspired from the field of complex networks is applied to a network associated with a SRS which is drawn by assuming that nodes are robots and links are informational connections with two nearest robots. This clustering method make it possible for us to find some subgroups of robots. Applying this method with the concept of behavioral sequence inspired from ethology, we are possibly to observe task allocation of a cooperative food foraging problem with a SRS contains 100 robots.

ABSTRACT. In nature, self-organization plays a role in making global and complex patterns by local interactions. In this paper, we purpose a distributed clustering control law for a one-dimensional robot swarm. The clustering control law uses a reaction-diffusion equation which models self-organization. We perform numerical simulation of one-dimensional robot swarm clustering.

ABSTRACT. Dung beetles show fascinating locomotion abilities. They can use their legs to not only walk but also manipulate objects. Furthermore, they can perform their leg movements at a proper frequency with respect to their biomechanical properties and quickly adapt the movements to deal with external perturbations. Understanding the principles of their biomechanics and neural computation and transferring them to artificial systems remain a grand challenge. According to this, we present here a first prototype of a real dung beetle-like leg developed by analyzing real dung beetle legs through micro-CT scans. We also apply adaptive neural control, based on a central pattern generator (CPG) circuit with synaptic plasticity, to autonomously generate a proper stepping frequency of the leg. The controller can also adapt the leg movement to deal with external perturbations within a few steps.

ABSTRACT. This paper proposes a shape control method for a snake robot maintaining the position and attitude of the robot's head. The snake robot has passive wheeled links and active joints. It can switch the allocation of grounded/lifted wheels using a rotation of joints, and has kinematic redundancy by introducing lifted wheels. We chose joint angles as controlled variables representative of redundancy, and directly controlled them. Sequential switching of lifted wheels from head to tail leads to a change in the robot's shape. Simulations demonstrate the effectiveness of the proposed method.

ABSTRACT. This paper describes a behavior acquisition of virtual robots by evolving artificial neural network (EANN) with a gradual learning. The gradual learning is a method in which initial states of simulation for evaluation is changing as optimization progresses. Motion of virtual robot is calculated by the physical engine PhysX, and it is controlled by an ANN. Parameters of an ANN are optimized by particle swarm optimization (PSO) so that a virtual robot follows the given target. Experimental results show that the gradual learning is better than a common learning method, realizing the standing behaviors which are not acquired by a common learning at all. It is also shown that random initialization of solutions in the middle of optimization leads to better behaviors.

ABSTRACT. Robotic systems inspired from humans have always been lightening up the curiosity of engineers and scientists. Of many challenges, human locomotion is a very difficult one where a number of different systems needs to interact in order to generate a correct and balanced pattern. To simulate the interaction of these systems, implementations with reflex-based or central pattern generator (CPG)-based controllers have been tested on bipedal robot systems. In this paper we will combine the two controller types, into a controller that works with both reflex and CPG signals. We use a reflex-based neural network to generate basic walking patterns of a dynamic bipedal walking robot (DACBOT) and then a CPG-based neural network to ensure robust walking behavior.

ABSTRACT. In this paper, we focus on peristaltic movements of the earthworm, and aim to develop the peristaltic mobile robot. To analyze the motions of the robots on the 2-dimensional flat surface including a turning motion, a dynamic equation of two-dimensional model is derived and numerical simulations are demonstrated.

ABSTRACT. Biological snakes have a string-like elongated body, and use their body to move forward. They can move in complex environment by adapting to the environment by changing its shape properly. They sometimes utilize irregularities, such as rocks and plants, to move faster and more efficient. This is called as obstacle aided locomotion. Obstacle aided locomotion motivate to develop a snake-like robot which can move in a challenging environment. In this paper, we describes a prototype of snake-like robot that moves by reflexible behavior. We have implemented moving methods to the prototype snake-like robot, that are lateral undulation, lateral inhibitioin and reverse lateral inhibition. Experiments are conducted by combining these behaviors in the environment where cyrindarical objects are arranged.

ABSTRACT. This paper presents a methodology to capture the heading angle of a Modular Snake Robot (MSR) executing lateral/helix rolling gaits. In order to determine such a model, an important part of this paper is devoted to describing the representation of the macroscopic spatial robot's motion in a world frame of reference. MSR motion is achieved by the coordinated and cyclic internal motion of its joints, which render an overall spatial motion. To decouple internal and external motion an algorithm that comprehends consecutive rotations of the robot's shape and frame transformations is applied. This algorithm provides a floating frame of reference that corrects the orientation of the robot's shape in real time referred from any body attached frame. Such frame of reference can serve as a basis to propose simple kinematic models. The success of the floating frame formulation for rolling gaits proposed here, relies on their analytic calculation using solely the already known control parameters needed to generate the gait instead of any regression or optimization method making fast and reliable.

ABSTRACT. A lot of types of snake-like robots and its motions have been studied. For example, a motion for a snake-like robot on a flat plane named lateral undulation was developed. However, in the motions, the robot's shape is planned in advance. On the other hand, snake-like robots cannot avoid contacting with the surrounding objects in an environment where many obstacles exist. In order to solve the problem, we proposed a method of obstacle aided locomotion and conducted some experiments by using a prototype of a snake-like robot. However, from our experimental results, it was found that the method is not always useful. In this paper, we describe simulation results of a snake-like robot in a narrow passage to check a relationship between robot's parameter and friction force to obstacles. We also propose a method to change behavior according to a direction of friction force between a snake-like robot and obstacles.

ABSTRACT. Animals like fire ants form entangled aggregates which can act like solids and fluids and whose properties are not understood. Inspired by these studies, we investigate a new class of active matter, granular media composed of smart particles, or “smarticles”. In our earlier work with static concave granular media, we discovered that ensembles of “u”-shaped particles can exhibit geometrically-induced cohesion by mechanically entangling via particle interpenetration. Furthermore, experiments indicated that the strength and/or extent of entanglement, and therefore cohesion could be tuned by varying only the outer link length, with respect to the center link, and the compaction of the ensemble. In this paper, we present a robotic granular particle, a candidate smarticle, which is a behavior and power autonomous 3-link robot that can be activated through audio communication. Since study of millions of such smarticles is cost and labor prohibitive, we also develop a simulated smarticle in the Chrono environment which is an open-source simulation engine for large scale multibody dynamics problem with impact and frictional contact. We envision that ensembles of such smarticles will enable us to systematically explore the physics of active granular media including issues of jamming, geometric cohesion, and glass-like dynamics.

ABSTRACT. Swarm robotics systems (SRS) are composed of many homogeneous autonomous robots without any form of centralized control. Evolutionary swarm robotics (ESR) has been considered as a promising approach in which robot controllers are represented by evolving an artificial neural network. But in ESR researchers have to confront a problem, that is impractically high conputational cost. On the other hand, MapReduce is a powerful framework for parallel computing proposed by Google. In this study, we describe how we design and implement ESR into the MapReduce framework on Hadoop, a widely used open source implementation of MapReduce. To demonstrate the efficiency and scalability of the proposed method, we evolve a SRS in the cooperative transport problem.

ABSTRACT. Self-organized flocking of robotic swarms has been investigated for approximately twenty years. Most studies 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, flocking performance depends on how these rules are configured and no guideline for the configuration exists. This paper investigates real robot flocking where individuals can switch their roles depending on the situations. Robots can move as leaders or followers, and the roles are dynamically allocated using stochastic learning automata. The flocking performance is evaluated in a scenario where robots consecutively travel between two landscapes.

ABSTRACT. In this research, we build a robot agent model and operating environments with easy configuration changes using three dimensional dynamics simulator, and attempt to analyze cooperative behaviours of robotic swarms. A robot agent is truly modeled based on an actual robot using a rapid prototyping method, and the simulation results of cooperative behaviour are compared with an actual robot's results.

ABSTRACT. This paper deals with the autonomous control of a biologically inspired autonomous ornithopter. An ornithoper is the category of micro aerial vehicles (MAVs) that provides lift and propulsion by the use of flapping wings in contrast to fixed-winged or rotary-winged UAVs. Basic system modelling and a description of the flying robot is first given. Then we discuss how feed-forward control can be implemented using multi dimensional parametric curves to specify simple flight trajectories. Subsequently, two disparate feedback control approaches are implemented for comparison. The first is in the form of a self-tuning linear adaptive controller that first involves acquiring a linear model of the dynamics being considered and then performs system identification. Using the identified parameters, the adaptive controller design is implemented using pole assignment. In the second approach, we describe ongoing works where instead of hypothesizing the system as linear, it is treated as non-linear and the latest non-linear, model-free controllers are designed.

ABSTRACT. Collective behavior of moving individuals in two-dimensional space is analyzed from the viewpoint of relative position with respect to their moving direction. It is clarified that a quantitative relation between spatial location and temporal delay is necessary to maintain the relative position. As an example, real data of flying hooded gull flocks obtained by stereo camera systems are analyzed.

ABSTRACT. We conducted field measurements of the trajectories of individuals in the skeins of wild geese. Using a stereo camera system, we estimated the operational distances between neighbors as well as response time of followers to leaders. We also analyzed occasional events in which single goose cuts into a skein of geese to become a member of the skein, which indicate that interactions are asymmetric; leading birds do not care about the followers. Furthermore, spatiotemporal analysis suggests that the trailing individuals loosely synchronize their wing beat in the way that the flapping oscillation of nearest neighbors tends to becomes in-phase at the same position in the flight direction. Based on the field measurements, we proposed a phenomenological equation of motion for collective flight of geese and discussed the conditions for stability of collective flight.

ABSTRACT. Honeycombs consist of a regular configuration of hexagonal holes. Although this self-organized and precise architecture has attracted much research, the rules by which it is constructed have not yet been elucidated. Here, we propose a simple and instinctive model for worker honeybees building a honeycomb. We assume that the wax cluster is dug linearly, guided by the workers' sense of touch. The regularity is not then produced intentionally; it is due to the competition between the growth of the wax and its excavation by the bees. We carried out two-dimensional simulations of our model, and obtained triangular and fishbone structures that are the triggers of the honeycomb construction process. Our study clarifies that anisotropic growth of the wax is a key factor in the construction.

ABSTRACT. The exploration-exploitation dilemma is a recurrent adaptive problem for humans as well as non-human animals. To solve this problem, humans gather information not only from individual try-and-error explorations, but also from many types of social ways, e.g., observing other individuals behavior, sharing 5-star ratings, transmitting word-of-mouths and so on. However, whether social information sharing might improve human decision performance is still largely unclear. Results from a laboratory experiment, in which human participants engaged an n-armed bandit task collectively, showed that enriching social information by allowing participants to share 5-star rating information had a detrimental impact on performance compared to the simpler information sharing situation. To shed some light on this empirical result, I also show the results from an individual based model that investigates how performance of collective decision-making is influenced by different types of advice-sharing systems. The relations between types of information sharing and collective intelligence under exploration-exploitation trade-off are discussed.

ABSTRACT. Cooperation is a hallmark of life on Earth, ranging from ancient self-replicating molecules to modern human societies, and has provided the driving force behind the major transitions in evolution [1]. In cooperative groups, public goods are usually generated and shared by component individuals at some personal cost. However, the publicness of the goods provides an incentive for some group members, called cheaters (defectors, exploiters, or free-riders), to benefit from the goods without contributing to them. The cheaters may ruin the public goods, or at least greatly diminish their utility. This situation, described as the “public goods dilemma” in game theory, makes the ubiquity of cooperation a major evolutionary conundrum [1]. Social insects, such as ants, bees, wasps, and termites, serve as a representative of cooperative animal groups. A defining characteristic of cooperation in social insects is reproductive division of labor, under which a reproductive role is limited to a subset of colony members (usually queens), with the other members (workers) concentrating on cooperative tasks. Mutant cheaters can undermine the reproductive division of labor by reproducing without helping. Recently we reported experimental evidence for the public goods dilemma in societies of the ant Pristomyrmex punctatus, which is known to have genetically determined cheaters [2,3]. In this species, the morphologically and functionally distinct queen caste is secondarily lost; instead, all morphologically defined workers are involved in both parthenogenetic (thelytokous) reproduction and cooperative tasks [2]. Reproductive division of labor is realized through age polyethism (or temporal castes) among workers, where younger workers reproduce and engage in inside-nest tasks such as brood care, and older workers cease reproduction and shift to outside-nest tasks such as foraging [2]. Cheater individuals, which constitute a single intraspecific lineage in the field [3], engage in few tasks except for reproduction [2]. Using experimental colonies with varying proportions of cheaters, we tested whether the fitness of cheaters and workers in a colony of P. punctatus conformed to the framework of the public goods dilemma [4]. We established 25 experimental colonies each consisting initially of 100 individuals with 0%, 25%, 50%, 75%, or 100% cheaters (5 colonies for each proportion), except for one colony which contained 73% instead of 75% of cheaters on account of a shortage of cheaters in the source colony. Colonies were kept in the laboratory for 64 days. After the rearing period, survival and brood production of individual cheaters were significantly greater than those of individual workers living with them (generalized linear mixed model [GLMM], likelihood-ratio test [LRT], p < 0.0001), whereas an increase in the initial proportion of cheaters significantly reduced the fitness of all adults (in both survival and brood production, GLMM, LRT, p < 0.0001). In this presentation we focus on the robustness and flexibility of the allocation of cooperative tasks. In social insects, social regulation of the division of labor often achieves remarkable robustness or homeostasis against both external and internal perturbations [5,6], attributable to the behavioral flexibility of colony members that respond directly or indirectly to the perturbations. The underlying mechanism has been a key issue in the study of self-organization in biological systems. In this study, we regarded the invasion of cheaters as another class of perturbation inherent in cooperative systems. Although it is well known that the reproductive division of labor is vulnerable to cheater invasion, less is known about other aspects of the division of labor in the face of cheaters, such as cooperative task allocation among workers. We measured the allocation of tasks to activity outside the nest, because it is a typical cooperative behavior shown by post-reproductive workers of P. punctatus. The proportions of individuals (workers + cheaters) found outside their nests, averaged over observation periods, was 3.8%–14.7% in the presence of workers (0%–75% cheaters), but only 0.4%–1.7% in the absence of workers (100% cheaters). When we introduced the presence/absence of workers (coded as 0/1) as an explanatory variable separate from the initial proportion of cheaters, the proportion of outside-nest individuals did not change significantly with the cheater proportion (GLMM, LRT, p = 0.1973), but the effect of worker absence was negative and significant (LRT, p < 0.0001). These results indicate that the outside-nest activity of the colonies remained stable as long as enough workers were present to carry out all tasks. The observed range is close to previous field data (mean ± SD = 5.9% ± 2.7% [7]). Moreover, a previous experimental study (excluding cheaters) showed that offspring production was maximized when 5 to 10% of the nestmates were post-reproductive workers who worked outside the nest [8]. These studies suggest the presence of regulatory mechanisms that maintain the optimum proportion of outside-nest individuals, thereby maintaining social resilience [9] that still works against cheater invasion. Previous studies identified age polyethism as a main cause of task switching in P. punctatus, whereby all workers shift from inside-nest to outside-nest tasks as they age [2], as corroborated by the present study where the proportion of outside-nest individuals increased significantly over time (effect of days from onset, LRT, p < 0.0001). Of special interest was who contributed to the outside-nest activity. In another GLMM, we included the individual’s affiliation (cheater/worker here coded as 0/1) as an explanatory variable. Workers had a significantly higher probability of engaging in outside-nest activity than cheaters (GLMM, LRT, Type II, p < 0.0001), and the outside-nest activity increased significantly over time (effect of days from onset, LRT, Type II, p < 0.0001). The analysis also revealed that the outside-nest activity of individuals increased significantly with the increase in the initial cheater proportion (LRT, Type II, p < 0.0001), and that workers contributed significantly more to the increase in the outside-nest activity than cheaters (interaction of cheater/worker × initial cheater proportion, LRT, Type II, p < 0.0001). Our results show that behavioral flexibility of workers has a place in task allocation of P. punctatus, as well as age polyethism, which manifested as an increase in the outside-nest activity over time. How workers can detect a shortage of outside-nest individuals is currently unknown, but the observed pattern suggests that the local density of individuals inside the nest, including cheaters, might be involved. Taken together, our results illustrate not only how a social dilemma created by interactions between heterogeneous system members affects the performance of self-organizing cooperative systems, but also how the self-organization shapes fitness consequences of both the members and the systems as a whole.

Keywords: social insects, public goods game, self-organization, social resilience, heterogeneous swarm

References

[1] J. Maynard Smith, and E, Szathmáry, The Major Transitions in Evolution, Oxford Univ Press, New York, 1995.

[2] K. Tsuji, and S. Dobata, “Social cancer and the biology of the clonal ant Pristomyrmex punctatus (Hymenoptera: Formicidae),” Myrmecological News vol. 15, pp. 91–99, 2011.

[3] S. Dobata, et al., “Cheater genotypes in the parthenogenetic ant Pristomyrmex punctatus,” Proceedings of the Royal Society B vol. 276, pp. 567–574, 2009.

[4] S. Dobata, and K. Tsuji, “Public goods dilemma in asexual ant societies,” Proceedings of the National Academy of Sciences of the United States of America, vol. 110, No. 40, pp. 16056–16060, 2013.

[5] C. Detrain C, J-L. Deneubourg, and J. M. Pasteels, Information Processing in Social Insects, Birkhauser, Basel 1999.

[6] S. Camazine, J. L. Deneubourg, N. R. Franks, J. Sneyd, G. Theraula, and E. Bonabeau, Self-organization in Biological Systems, NJ: Prince. Princeton, 2001.

[7] K. Tsuji, “Reproductive conflicts and levels of selection in the parthenogenetic ant, Pristomyrmex pungens: contextual analysis and partitioning of covariance,” The American Naturalist vol. 146, pp. 586–607, 1995.

[8] K. Tsuji, “Inter-colonial selection for the maintenance of cooperative breeding in the ant Pristomyrmex pungens: a laboratory experiment,” Behavioral Ecology and Sociobiology vol. 35, pp. 109–113, 1994.

[9] A. B. Sendova-Franks, and N. R. Franks, “Social resilience in individual worker ants and its role in division of labour,” Proceedings of the Royal Society B vol. 256, pp. 305–309, 1994.

ABSTRACT. In this paper we propose a model of righting behavior of a starfish. Starfish have two layers of motor organs of arms and tube feet. Adaptive righting behavior is reproduced by implementing interaction between these layers that was more efficient than by our previous model.

ABSTRACT. Collective motion of flagella and cilia is essential for controlling the swimming of cells. We introduce a simple rotor model and study the roles of hydrodynamic interaction between the active filaments, focusing on (i) collective flow in a bacterial carpet and (ii) metachronal waves of cilia. For each system, we obtain a rich phase diagram of collective behaviors that are characterized by the long-range and tensorial nature of the interaction.

ABSTRACT. Animals must move to search for other organisms which can be a mating partner, prey or members of the same group. Over the last years, the interest has grown in understanding the drivers of random search efficiency [1]. In a pure random search, the organisms are assumed to look for targets without ant prior knowledge about their possible locations, so the only way in which the target detection rate can be effectively enhanced is by avoiding already visited places. Some studies have measured movement patterns of various organisms qualitatively and have theoretically estimated optimal searching strategy under several conditions [1]. However, as most of the study on optimal search strategy focused on foraging for food, it is controversial whether other searching activities such as for sexual partners are optimized in terms of their efficiency [2]. In searching for mating partners, as the movements of targets affect the search efficiency of the searcher [3], male movement patterns should be most encounterable to female and vice versa. In this study, as a first step, we investigate the movement patterns of males and females of termite dealates (wing-shed adults) by applying a novel technology for tracking small insects. Termite colonies generally are founded by dispersing winged primary reproductives. In a certain season, alates (winged adults) all fly off at the same time in great swarms, and their flight orientation is aimless and waving [4]. After dispersing, alates break off their wings and each dealate runs for a few days until they encounter a dealate of the other sex with little or no information on where other individuals are located. Once they found their new nests, the dealates never forage outside. Therefore, their movements could be considered as a search for mating partners. On this point, termite dealates provide a unique opportunity for studying sexual difference in movement patterns. During a swarm season, live alates of the Formosan subterranean termite Coptotermes formosans were collected using light traps in Gobo, Wakayama, Japan. We chose males and females that shed their wings (N = 5 for each sex), and all experiments were conducted within 24 hours after collection. One of the ways to measure walking trajectories of individuals is allowing individuals to walk freely in closed arena such as Petri dish or container. However, in closed arena, we cannot detect a straight displacement like ballistic movement beyond the diameter of the closed arena. To solve this problem, we used an omnidirectional treadmill mechanism for small insects. This mechanism is originally configured as a cockpit of the insect-controlled robot, named ANTAM (ANT Activates Machine) [5]. Not to give a bias arising from light direction, we covered the cockpit by a PVC pipe covered with aluminum foil, with LED light coming from upside (Fig. 1a). Before each trial, we wiped the cockpit using 70% ethanol to remove possible effects of pheromones. The movement of each termite was tracked for 30 minutes on the cockpit with a frequency of 20.23 ± 2.79 frames per second (FPS) (mean ± sd). As the frame rates were uneven, we reduced the frame rate to 2 FPS for the following analysis. We discarded data of the first 10 minutes to avoid movements affected by manipulation. We obtained a trajectory of 20 minutes for each termite. We computed the instantaneous speed of each frame as a virtual 2-dimensional distance covered by a termite from the frame to the next. We successfully obtained walking trajectories of all the termites (Fig. 1b). From visual inspection, it could be inferred that males and females showed different patterns of movement. In comparison to females, males seemed to behave diffusively. To confirm this, we computed the Euclidean distance (d) from the beginning point of the trajectory for each frame. Then we used liner mixed models (LMM) to test for the effects of moved distance (Z-score transformation) and interaction of sex and moved distance on d, where the effect of variation among individuals was treated as a random effect (random intercept). This revealed a significant effect of sex on the increase rate of d according to increase of moved distance (as evidenced by a significant effect of the interaction between moved distance and sex; χ2 = 11023, p < 0.001). On the other hand, we found no significant effect of sex on the instantaneous speed of termites (LMM, χ2 = 0.334, p = 0.5633). All analyses were conducted using the “lme4” packages implemented in R v3.1.2 software [6]. The termite dealates showed sex-specific movement patterns, that is, males behaved more diffusively than females. In our experiments, all termites moved under the same environmental conditions and had no information about the location of targets. Therefore, the sex difference in movement patterns may reflect the sexual dimorphism in inherent characteristics of random search. Sexual difference in movement patterns is closely related to the sex pheromone excreted by females to attract males [7]. In some species of termites, it has been known that females attain a calling posture by raising their abdomen, where males are attracted to a calling female and touch the latter with their antenna and mouth parts [8]. However, sex pheromone in C. formosans is not such a volatile pheromone but a non-volatile pheromone [9]. This may implicate that this non-volatile pheromone is sufficient for enhancing the encounter rate between males and females like a kind of parasitoid wasp [10]. In this study, we have succeeded in a measuring of termites walking trajectory. The next step will be to investigate whether the observed sex specific pattern is optimized for efficiency of encounter rate and how this difference is driven by natural selection. By focusing on broad aspects of searching activity which can elicit interactions other than prey-predator interactions, we will be able to understand the movement patterns of organisms more deeply and mimic optimization of efficiency in random search under various circumstances.

ABSTRACT. Open-field tests are one of the most famous techniques to study locomotor activity to classify the effect of drug, gene or psychiatric diseases in rodents. We studied the effects of apparatus size focusing on walking durations. We explored the open-field behavior of mice utilizing circular open fields of various diameters. When the diameter of the test apparatus was greater than 75 cm, the walking period duration followed a three-exponential Poisson distribution function. We found that the properties of the scaling exponents and model selection became similar for test apparatus diameters greater than 75 cm. Based on the results, we discussed the mechanisms utilizing a state change stochastic model with walking speed of mice in open field.

ABSTRACT. A situation-dependent switching of the strategy of errors in group foraging of ants is studied. The strategy of errors is a stochasticity-associated foraging tactics of ants proposed by Deneubourg(1983), which indicates that errors of individual ants in following trail pheromone increases the efficiency of the pheromone-mediated group foraging. In this study, we attempt i)to extend Deneubourg’s original idea of the strategy of errors to incorporate the non-uniform distribution of errors in following pheromone trail by foraging ants, and ii)to show that a switching takes place between alternative optimal strategies of errors depending on foraging environment.

ABSTRACT. In this paper, we proposed a simple system that could control the direction of a swarm of mobile robots in two dimensions by using phase gradient. We implemented a system using a network of oscillator in which each robot was equipped an oscillator with local interaction ability. By manipulating parameters of an oscillator in a robot, we realized a particular phase distribution that controls the direction movement of the swarm robots. We verified the proposed system by using modified e-puck robots.

ABSTRACT. Quadrupeds change their gait patterns in response to locomotion speed, body properties, and environment. Although several hypotheses for the gait transition mechanism have been proposed, they could not fully explain the transition among versatile gait patterns. In this study, we propose a new hypothesis for the mechanism of quadruped gait transition using a simple model of embodied coupled oscillators. The proposed model is validated via simulation. We can conjecture from our results that three factors, 1) body support, 2) constraint of legs in the stance phase, and 3) interaction between fore and hind legs via body trunk, are essential for the quadruped gait transition.

ABSTRACT. Autonomous decentralized control based on mechanical interaction could be the key to develop robots that move adaptively and resiliently like real animals. However, systematic method for designing it has not yet been established. To tackle this problem, we focused on peristaltic locomotion of earthworms as our model, and proposed a decentralized control scheme in which "TEGOTAE", a Japanese word meaning reaction after the generation of some action, was fully exploited. We confirmed the validity of the proposed control scheme via simulation.

ABSTRACT. The purpose of the research is to establish multi-robots system, which consists of numerous numbers of mobile robots. To realize a flocking behavior with such multi-robot system, it is important to achieve a localization of each robot. However, it is difficult to apply the conventional localization method for such multi-robot because of a cost and computational resource. In this paper, a new localization algorithm for multi-robot system based on an ad-hock network of robot is proposed. And we apply the proposed localization method for the realization of some flocking behaviors of multi-robot system.

ABSTRACT. The paper deals with a leader-follower flocking control for a multi-agent system which consists of multiple agents and a leader. The proposed method enables all agents to be disposed at an uniform distance, reach agreement on a same velocity with their neighborhood, and to track a moving leader independently. Leader is seen as one of the agents, in addition, all agents obtain their goal using their neighborhood information only. Furthermore, we suggest the bounded potential function to guarantee the collision avoidance and connectivity maintenance between the agents. As a result, all of the swarm can move with preserving flocking configuration. Finally, numerical simulations show the effectiveness of the proposed method.

ABSTRACT. This paper studies robustness of synchronization for high-order heterogeneous multi-agent systems. Using the concept of the averaged dynamics, it is claimed that two sources enhance the robustness of synchronized behavior against heterogeneity of agents' dynamics. First, strongly coupled network makes each behavior of heterogeneous agent close to that of the averaged dynamics. Second, we observe that the averaged dynamics becomes more robust against individual variation as the number of agents increases. Simulation result confirms our claim that a large number of agents with strong coupling have robust synchronization.

ABSTRACT. This paper discusses a target enclosing behavior when a part of the orbit is blocked. Usually, an orbit to enclose a target supposes to be clear. However, this situation often takes place in realistic environment. For this issue we propose an adaptive method to enclose a target. The group of robots changes their direction according to the blocked part. For switching their direction Vicsek's work is applied

ABSTRACT. We know concurrent learning problem in multi-agent learning. In this paper, we show an essential environment including the problem. Furthermore we propose a solution for the problem in which a secondary system that is proposed in our previous paper is combined with Profit Sharing. The effectiveness of the proposed system is confirmed by numerical experiments.

ABSTRACT. At multiple spatial scales humans can act as teams, dyads and individuals solve problems by executing adaptive and predictive movements. In addition to the individual intelligence, collective intelligence has been demonstrated that they can have either positive or negative effects on the group; however, the intelligent behavior remains unknown at multiple spatial scales. The real-time mobile and cooperative groups, such as those where the hierarchically goal-directed individuals make specific decisions as a team may exhibit apparently complicated behaviors. Sophisticated teams in ballgames are excellent examples of how individuals can make different decisions separately to solve a hierarchical problem at team, dyad, and individual levels. The aim of this study was to investigate the real-time group problem-solving in multiple spatial scales during a ballgame. We focused on the basketball team performing person-to-person defense against an offensive group. Results demonstrated that the attacker-defender distance at minimum (1-on-1) and global (5-on-5) scales adjusted by spatial and predictive factors explained success and failure of the team outcome (i.e., shot) rather than the actual Euclidean distance. In meso-spatial scale (2-on-2), more than half of the shots were related to the screens to disrupt defensive players by legally blocking them. During the screens in an individual scale, three patterns of helping behaviors toward the teammate suffering the screen were observed: (1) in a hazardous situation, the agent abruptly helped the teammate to switch in the roles each other; (2) if not, the agent left the teammates or (3) transiently helped the teammate and quickly returned to handle the problem.

ABSTRACT. One approach in swarm robotics is homogeneous system which is embedded with sensing, computing, mobile and communication components. In this study, a target detection problem, which is one of navigation problems, was employed. Once a robot detects a target, robots immediately communicate with a base station via intermediate relay robots due to the multi-hop transmission of wireless communication. Therefore, this control task is completed with connectivity of the network. In a target detection problem, we must improve the performance of exploration as well as connectivity of the network. This study investigates the performances of the two kinds of random walk algorithm in navigation while loosely ensuring connectivity of the robotic network based on our previous study.

ABSTRACT. This paper proposes a new response threshold model to implement an autonomous specialization for swarm robotics. The response threshold model is a computation model of the sensitivity of ants in regard to external stimuli. It is considered that an ant can specialize either a worker or a non-worker autonomously because an ant has the different sensitivity in regard to external stimuli. The conventional response threshold model has used as an external stimulus the ratio of workers in a colony. However, an ant cannot know the state of all ants because ants can only communicate with neighborhood ants through pheromones. Therefore, the proposed response threshold model uses as an external stimulus the ratio of workers who an ant touches during short term. In this paper, we investigate the efficiency of the proposed response threshold model through the ant foraging problem simulations. As a result, we verified that the ants could collect foods effectively by switching either a worker or a non-worker statistically by using the proposed response threshold model.

ABSTRACT. Conditions of group foraging were experimentally constructed in an I-shaped maze using newly hatched domestic chicks as subjects. As predicted by the producer-scrounger game, the paired condition significantly enhanced the foraging efforts, and caused a precise matching in their feeder utilizing time indicative of a synergism. Lesion experiments further revealed that distinct populations of the descending projection neurons in arcopallium (major efferent isocortical area) are specifically involved in the facilitation. Visual attention to conspecific individuals could thus be critically involved in the social facilitation and behavioral synchronization of foraging flock of domestic chicks.

ABSTRACT. We investigate time development of the solid torus shape of fish schools by analyzing the moving images filmed from the bottom of a tank of an aquarium. Through using automatic measurement velocity vector data by means of method called optical flow, we review the effectiveness of the analysis based on rotation curve, namely distance from the center in relation to rotation velocity. Collective motion of the fish changes non-stationarity and continuously by several seconds. Yet, if we average the rotation curves by a dynamical time required for each circling movement in torus, universal structure of rotation curve appears.

ABSTRACT. We show that cooperative breeders, especially wasps and bees, exhibit greater productivity than lone breeders and show that this high productivity is achieved by reducing the number of broods to breeder ratio. We first formulate and discuss stochastic models for the theoretical productivities of both lone and cooperative breeding. Using hover wasp data, we then investigate the productivities under various broods to breeder ratio and conclude that cooperative breeders rear each brood with high quality care.

ABSTRACT. We analyse the total payoff of cooperative foraging with insurance scheme when the instantaneous payoff is linear to the group size. Cooperative foraging by primitively eusocial insects is paradoxical because the payoff is reported to be linear to the group size and no apparent advantage for the cooperation. We use Markov decision process to model the payoff of cooperative foraging with insurance scheme which is realised by remaining foragers who take care the brood of died foragers. Despite to the additional cost shared by remaining foragers, we find that in the low-risk environment, the insurance works effectively among the cooperative foragers, and achieve higher total payoff.