SWARM2017: THE SECOND INTERNATIONAL SYMPOSIUM ON SWARM BEHAVIOR AND BIO-INSPIRED ROBOTICS
PROGRAM FOR WEDNESDAY, NOVEMBER 1ST
Days:
previous day
all days

View: session overviewtalk overview

09:30-10:30 Session WeAa: Keynote speech (3)
Location: Hall (Rooms I & II)
09:30
Biologically Inspired Snake-like Robots and their Practical Applications

ABSTRACT. In 1993, I published the book entitled “Biologically Inspired Robots Snake-Like Locomotors and Manipulators-” from Oxford University Press. It was my secret pleasure to know that the term “Biologically Inspired Robots” has become very popular since this time in Robotic community, but at the same time it was a little bit disappointing for me to know that very few of them were really used in real applications. In this talk, I will show my early study about the biological experiments using real snake and study of snake-like locomotors and manipulators, including world first snake-like locomotor ACM 3. At the same time, I will also explain that the biologically inspired snake-like robots were just recently really used in real applications, such as the 4 m long coupled-tendon snake-like manipulator CT-Arm used for the inspection of high radiation site of Fukushima Daiichi nuclear reactor accident, and multi-wheeled snake-like locomotor ACM R4 for the same purpose and to look in the burrow of Wombat.

10:30-11:30 Session WeAb: Invited talk (2)
Location: Hall (Rooms I & II)
10:30
Inverse Bayesian Inference in a Swarm of Soldier Crab

ABSTRACT. Although animals showing collective behavior have a tendency of getting together, they sometimes avoid their group mates to make a swarm being split. Through the experiments on behavior of soldier crabs, we confirm that soldier crabs’ decision making on whether they follow mates or not is dependent on local and temporal condition, and that swarming behavior equipped with intrinsic diverse motions could be explained by mutual anticipation under asynchronous updating. Recent study clarify that some social animals could make decisions based on Bayesian inference. The natural conditions surrounding animals are, however, instable, and optimal strategy chosen by Bayesian inference is inadequate and insufficient. We here introduce not only Bayesian but inverse Bayesian inference to mimic soldier crabs’ decision making. Inverse Bayesian inference is proposed by one of authors and is implemented by changing hypotheses due to the environment. We show that collective behavior of soldier crabs could be explained by a pair of Bayesian and inverse Bayesian inference, and that mutual anticipation with asynchronous updating could be expressed as Bayesian and inverse Bayesian inference with synchronous updating.

12:30-13:30 Session WeP1a: OS2: Insect's movement patterns revealed by virtual reality system (1)
Location: Room I
12:30
Measurement System Based on Robotics for Studies of Insect Behavior

ABSTRACT. Animals routinely search for food, mating partners, nests, etc. in their lives. Their behavioral patterns of search must be sophisticated in the process of biological evolution. On the other hand, engineers have studied on optimal searching algorithms for targets with arbitrary distributions or movements in solution space. One of the solutions for exploring searching algorithms is the engineering applications of behavioral patterns of animals, which requires the techniques to measure their behaviors in details. In the field studies, it is difficult to measure the whole behavioral process of animals because researchers need to chase animals for several kilometers to track their movements. In laboratory conditions, image-based tracking is widely employed to analyze searching behaviors of relatively small animals. However space constraints often becomes a problem to observe natural behaviors of animals because of the use of wall-bounded arenas (e.g., petri dishes) in the laboratory. In order to solve this problem, the following two methods to provide an infinite two-dimensional space with animals have been developed; 1) an un-driven treadmill with a fixed animal which drives a lightweight sphere, 2) a treadmill with an un-fixed animal on a sphere driven by servomotors, and they have advantages and disadvantages. In the present study, we have developed a servosphere for the robot named as ANTAM which is steered by insect as a pilot. To develop a robotic systems steered by insects, we focused on the following required designs; i) not disturbing insect behaviors, ii) insects not escaping from the cockpit. Thus, the ANTAM was developed as a system for measuring trajectories of insect movements. In this presentation, we will explain the background of ANTAM development and present possible methods for measuring insect behaviors.

12:50
Improvement of ANTAM for long-term measurement of small animal movements

ABSTRACT. It is important in behavioral ecological research to simultaneously measure the environmental situation and behavior on how the organism adapts and acts in its environment. When insects are observed, there are two approaches to measure them, roughly divided into behavior measurement in a field study and in a laboratory experiments. Behavior measurement performed in the field can acquire behavior trajectory data in an actual environment by using sensors such as GPS etc. However, since it is affected by external factors such as enemies, wind, temperature, in order to clarify the causal relationship, a sensor system for accurately measuring the external environment is required. Meanwhile, in the laboratory can measure in a under controlled environment, but it is mainly conducted in a limited and small space such as a petri dish, so the measured data are difficult to identify with in the field. For this reason, several studies including our research have developed experimental apparatuses for measuring behavior in environments where there is no boundary that restricts behavior. In this study, we clarified the problem of a previous tracking system named ANTAM in experiment and evaluated after improvement both of software and hardware. As a result of improvement, it was confirmed that the multithreading control program is more stable than the single thread. Also, by using xvid as a video codec, long time recording over 5 hours became possible. Furthermore, by improving the data acquisition method from the mouse sensor, it became possible to acquire the movement amount at the sampling rate of 125 Hz, so the behavior trajectory was more precise.

13:10
Sexually differential movement can enhance mating encounters: potential uses of servosphere for detecting intraspecific variations of walking patterns.

ABSTRACT. Animals face spatially and temporally complex structured environments to search for food, hosts and mates. When individuals have little to no information about target locations, selection acts on the search strategy efficiency. How should they move to search for targets most efficiently? This is called the random search problem, which is relevant across fields [1]. One leading hypothesis is Lévy walk search hypothesis [2]. The Lévy walk is a special class of random walk models in which the probability density function of step length l has a power-law tail: (1 < μ ≤ 3), where μ is a power-law exponent. Lévy walks have attracted a great deal of attention because it has advantage among random search strategies to encounter targets that are randomly and sparsely distributed [2-3] and has been observed across species of animals [4-7]. Lévy walk includes a variety of random walk processes from ballistic motion (μ → 1) to Brownian random walk (μ > 3). In spite of this diversity, previous studies have mainly focused on whether Lévy walk characteristics presents in movement patterns of animals or not, resulting that little is known about a variation of movement patterns among or within species. As different movement patterns can result in different searching efficiency and fitness, exploring the maintenance mechanism of this variation will lead to our better understanding of the evolutionary optimal search strategies or the movement patterns that animals follow. One of the fundamental variations within species is sexual difference. Animals have evolved various sex-specific characteristics to improve the efficiency of encounters for successful mating. In mate search, both sexes can mutually optimize their movement patterns to enhance encounter rates. Although previous theoretical studies on the search efficiency of Lévy walks have acknowledged that foraging can be for arbitrary types of targets (food, mates, or habitats), not a few of them have assumed, explicitly or implicitly, that foraging is for food; they have assessed how many prey items a predator can find [2-3, 8 but see 9]. In this study, we first examined the effect of sexual difference of movement patterns on the mating encounter rates (details in published work [10]). We developed a mutual search model, where n individuals with equal numbers of males and females (nm and nf) move in a two-dimensional borderless space and search for individuals of the opposite sex. Males and females search for each other by performing a Lévy walk with sex specific exponent (male: μm, female: μf) at a velocity of v until encountering another individual of the other sex. All individuals have spherical bodies with the same diameter Φ. When the distance between the centers of a male and a female becomes smaller than Φ, the two individuals encounter one another, where we did not consider the encounter of two same-sex individuals. On encountering one another, the pair stops searching and disappears from the searching space. Searching time lasts for tmax. All individuals are randomly distributed on a square grid, separated by at least d at the initial condition of each generation. As a result, we found that individuals in a sexually dimorphic population can achieve the highest fitness when the lifespan was intermediate. Although the optimal combinations of μm and μf varied depending on the population size and the lifespan, sexually dimorphic movement patterns enhanced the mating encounter rate with a certain lifespan irrespective of the population size. Then, we proceeded the evolutionary simulations to investigate whether selection acting on the efficiency of encounters can drive the evolution of differential movement patterns between the sexes. We expand the above simulations by changing the searching time from tmax to until the proportion r of individuals encountered. The value of r can be regarded as a selection strength. We considered the power-law exponent μ for Lévy walk as an evolving genetic trait. All individuals have their own μ values that are inherited by their same-sex offspring with mutations described below. In other words, in our simulations we assumed haploid genetics without recombination, whereby the gene determining μ links to the respective male- and female-determining loci. Mutations take place during the formation of next-generation individuals at a rate of 0.001, where the μ value increases or decreases by 0.1 between 1.1 and 3.0. The initial μ value, μ0, was set at 3.0 for all individuals. Our results show that dimorphic movement patterns evolve under an intermediate r and sufficiently large population size. As population size increases, the strength of selection r with which dimorphic movement patterns evolve also slightly increases. This is because, with a large number of individuals, the risk of moving in a direction without mates declines and thus the rate of encounter of highly diffusive walkers improves. We provided a novel mechanism for the evolution of sexual dimorphism. Then how we will be able to test this theoretical prediction using the real organisms. A potential candidate could be found in subterranean termites. After newly emerged adult termites fly off to disperse from their natal colonies, they shed their wings and run to search for mates without food and locational information; the resulting pair then becomes the royals and maintains lifelong monogamy [11]. In this process, termite individuals must be under the necessity of only search for mates without locational information. Nevertheless walking patterns of termite individuals have never been explored because of its technical difficulties. One of the ways to measure walking trajectories of small 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. Recently, we developed a system named ANTAM by applying an omnidirectional treadmill mechanism, or a servosphere, as a tool for tracking two-dimensional movements of small animals that can provide both a homogenous environment and a virtual infinite space for walking [12]. This system enables us to measure spontaneous movement patterns of small individuals for theoretically infinite time. On our system, animals showed similar diffusive characteristics to those observed in the large arena simulating an open space. Furthermore, we showed that anomalous diffusion properties, including Lévy walk, can be detected from the free-walking behavior on our tracking system. On the other hand, moving speed and turning angles are different among the tracking conditions, suggesting that our system may not be suitable to measure such characteristics. However one primary advantage of our system is to detect the ecological meaningful differences in spontaneous movement patterns such as the sex differences. In conclusion, we have proposed that a novel approach to focus on the inter- and intra- specific variation of the movement patterns. As an example, we have shown that the optimization of mating encounter can result in sexually dimorphic movement patterns. The variation of spontaneous movement patterns (including sex difference) can be detected by using ANTAM. Measuring movement patterns focusing on the various ecologically significant aspects of the organisms (sex, physiology, lineage, age, etc.) will open up a new direction in the study of movement patterns.

12:50-13:30 Session WeP2a: Bio-inspired studies (2)
Location: Room II
12:50
Constructions of Strict Lyapunov Functions for Homogeneous Time-Varying Systems with Applications in Multi-Agent Systems

ABSTRACT. We provide a new technique for constructing strict Lyapunov functions for time-varying systems. The definition of homogeneous auxiliary system is given, where it is assumed that certain homogeneous functions are admitted with their derivatives, in terms of the error systems, bounded by periodic functions. Based on the homogeneous auxiliary system, sufficient conditions of uniform asymptotical stability for time-varying systems are formulated using the homogeneity framework. Unlike classical constructions of strict Lyapunov functions, where non-strict Lyapunov functions or persistency of excitation condition are required, our criterion is greatly relaxed for broad classes of systems. The utility of our result is illustrated through the well-known consensus problem of multi-agent systems.

13:10
Distributed Consensus Control of Multi-agent Systems: A Mixed H2/H∞ Approach

ABSTRACT. This paper is devoted to consensus problems of undirected topology multi-agent systems with external disturbances. New class of feedback H2/H∞ control protocols are proposed to reject the influence of the external disturbances to the multi-agent systems. These protocols not only can bound the linear quadratic of consensus, but also can guarantee the H∞ performance of the multi-agent consensus. By utilizing H2/H∞ convex sub-optimization approach, sufficient conditions of reaching H2/H∞ consensus index are proposed in the form of linear matrix inequalities (LMIs).

13:40-14:40 Session WeP1b: OS2: Insect's movement patterns revealed by virtual reality system (2)
Location: Room I
13:40
Try the Measurement of Walking Activity by ANTAM in the Beetle Strains Selected for the Duration of Death-feigning

ABSTRACT. The duration of death-feigning, i.e., thanatosis, has been diverged to shorter and longer directions in the red flour beetle, Tribolium castaneum (Herbst) by artificial selection experiment (Miyatake et al. 2008). Beetles derived from strains selected for shorter duration of death-feigning showed higher expression of dopamine in their brain and higher activity than these derived from strains selected for longer duration of death-feigning. In the present study, we tried to measure walking activity by a system named ANTAM (Nagaya et al. in prep.) by the Kramer-treadmill systems. In the present session, we will compare the results of walking distance measured by ANTAM and another measuring system. Also, we will report for some points concerning minor modifications of the present ANTAM to measure walking behavior of the beetles.

14:00
Pattern of Ant walk – Learning from Ant-mimicking spider

ABSTRACT. The jumping spider genus Myrmarachne (Salticidae) consists of over 200 species of morphologically accurate ant mimics, enabling Myrmarachne to evade ant-averse predators that confuse the spiders with ants (Batesian mimicry). The ant-like appearance of Myrmarachne is due to a narrow, constricted abdomen and cephalothorax, which creates the illusion of ant body, possessing three body parts (head, thorax and abdomen) rather than two (cephalothorax and abdomen), as found in spiders. In addition to morphological mimicry, Myrmarachne is characterized by behavioral mimicry. For example, Myrmarachne raise and wave their front pair of legs about in a manner similar to the movement of ant antennae, and walk on 6 legs like ant (Ceccarelli 2008). On the other hand, non-mimicking jumping spiders use 8 legs for walking, and rarely run, instead usually moving with a ‘choppy’ movement pattern with frequent changes of direction (Jackson 1990). The aim of this study was to determine whether the locomotion of the Myrmarachne ant-mimicking jumping spiders closely matched to its ant models and to identify examples of species-specific locomotory patterns within the spiders and their model ants. We quantified the locomotory pattern of several species of non-mimicking jumping spiders, Myrmarachne, and ants from tropical forest of Borneo and Thailand, employing computer tracking analysis. Here, we present that the locomotion of Myrmarachne not only mimics generally resembles that of ants, but also species-specifically matched to its ant-model species. Our results indicated that there is selection pressure on Myrmarachne to detect and copy ant characteristic motion for increased resemblance to a model. In this aspect, it can be said that ant-mimicking jumping spiders have much to teach us in the design of robots which mimic ant behavior.

References [1] Ceccarelli, Fadia Sara. "Behavioral mimicry in Myrmarachne species (Araneae, Salticidae) from North Queensland, Australia." Journal of Arachnology 36.2 (2008): 344-351. [2] Jackson, Robert R. "Predator‐prey interactions between jumping spiders (Araneae, Salticidae) and Phokus phalangioides (Araneae, Pholcidae)." Journal of Zoology 220.4 (1990): 553-559.

14:20
Pill bugs' behavioral patterns revealed in virtually infinite multiple T-maze

ABSTRACT. Pill bugs (Armadillidium vulgare) demonstrate a behavior called turn alternation that keeps their overall direction of movement straight after obstacles in experimental settings force them to deviate from a course. For example, this behavior is seen when they alternate their path choice on successive trials of the T-maze test. However, sometimes pill bugs stop after turning and change their direction (directional change). The function of this directional change has not been investigated because such individuals are usually omitted from the data. The present paper shows that pill bugs use directional changes to prevent them from turning in the same direction on two successive turns, a behavior called turn repetition. We examined the behavior of 36 pill bugs that each completed 130 successive T-maze trials. Directional changes appeared more frequently when individuals had begun a turn repetition than when they had begun a turn alternation. Furthermore, after correcting for turn repetition, turn alternations increased.These results suggest that pill bugs have an inherent mechanism that acts to maintain turn-alternating behavior.

13:40-14:40 Session WeP2b: Swarm science (2)
Chair:
Location: Room II
13:40
Circular regression models for identifying abnormal parts in swarm behaviors and their quantitative characterization: data science approach

ABSTRACT. Extending circular-circular regression, we developed the statistical models that quantitatively characterize rotating school of fishes and their temporal changes. The methodology allows us to capture specific parts that exhibit abnormal movements in a swarm and to evaluate abnormal aspects and their degree. The number of abnormal parts are determined by model selection using widely applicable Bayesian information criterion (WBIC). We demonstrated our methodology for a rotating school of sardines in an aquarium.

14:00
Memory use in nest relocation of Diacamma sp. from Japan

ABSTRACT. Groups of individuals may relocate to another site for a variety of reasons [1–4], for example due to a change in the environment or a disturbance (e.g., predators or from anthropomorphic origin [5,6]). Studies have been interested in elucidating how consensus, allowing for whole groups to relocate to the same site, is reached, how collective fleeing is affected by group size, and how stable can a site choice be when faced with disturbances [3,7–9]. For instance, studies on numerous species have shown how social learning, indirect conspecific cues, and quorum-based dynamics can lead to a consensus [8,10–13]. In addition, collective choices have been classically studied focusing on the effects of direct (presence of conspecifics) and indirect (chemical markings) social interactions [10,14–16]. Recently, however, research has started to focus on the effect of individuals (their personality and memory) on the final, collective, choice [17–19]. Indeed, recent research has found, for example, that individual memory during foraging behavior is enough to trap all the foragers in a energetically poorer food source, similarly to effects conventionally attributed to established pheromone trail becoming too strong for new trails to compete [20]; on the other hand, it has been also shown that individual memory and experience can improve the collective performance [21]. Furthermore, in tandem-running ants, studies have shown that follower ants learn from the tandem leader the position of food sources. In turn these tandem followers become tandem leaders and thus the lessons are transferred, making tandem runs a slow process that nonetheless propagate time-saving knowledge among individuals [22]. Recently, relocation dynamics has become an important interdisciplinary topic that combines different concepts such as decentralized control, self-organization, collective decision making and social evolution, and is studied by combining new technological skills (RFID, Video tracking) and classical methods (capture-release-recapture). Social insects have become the central biological models for such studies. Indeed, such interest in social insects and group relocation is understandable since it is probably one of the most difficult collective choices as it results in the relocation of whole societies [23]. In tandem-running species, the consensus reached for a relocation nest is the result of the threshold number of nest-mates at the new nest and direct comparisons by tandem leaders of different nest-sites. These direct comparisons constitute a very interesting phenomenon in collective behavior because they might provide a mechanism to minimize the risk that an inferior nest-site is chosen or a better site is overlooked due to erroneous information [23]. Indeed, consensus reached solely on social information, where individuals choose a shelter where most individuals are located (e.g., cockroaches [8]) can sometimes lead to choosing a less optimal shelter (but have the advantages of being in a group: protection against desiccation, predators, etc.). Following this scientific trend in studying colony behavior both at the colony and individual level, we study the relocation dynamics of the queen-less ant Diacamma sp. from Japan and focus on the role of the memory of tandem leaders in choosing a new nest. Diacamma belongs to the subfamily ponerinae and is distributed in Asia and Oceania, from eastern India to Japan, and from southern China to northeast Australia. They form colonies of medium to small size (around 100 workers), and readily relocate nest site after disturbances [24,25]. The genus Diacamma uses tandem runs when relocating to a new nest: A forager will lead single worker towards the new location. During a relocation event, the foragers act as leaders, and among those leader there is generally a single individual that is more active in the relocation (initiates more tandem runs) [29,30]. However, any leader lost, including the principal tandem leader, are eventually replaced [31]. The experiments we carried out were as follows, a Diacamma colony was put in a controlled environment with two other nest choices. During the first stage these nests choices were available to the ants and could be explored. During the second stage their colony was disturbed (nest is left to desiccate and illumination inside was increased), which led them to relocate into one of the two nest choices. Once they had relocated, they were placed in their original nest and the experiment repeated, except this time their previous choice was inaccessible to them before being disturbed (Stage 3), and only rendered accessible before they began relocating (Stage 4). The aim of this study was to elucidate how the memory of tandem leaders affected the relocation dynamics.

14:20
Universality of Individual Movement in Ant Colonies

ABSTRACT. Animal behaviour occurs in bouts interspersed with periods of inactivity. One manifestation of this is the intermittent locomotion of animals [1]. Bouts of activity and inactivity occur in the absence of any environmental stimuli [2] but such a spontaneous stop-and-go signal has to adapt to external conditions, including the social environment. Two important questions from the point of view of both collective behavior and swarm robotics are: (1) How do individuals decide when to stop and when to go? and (2) How does the number of interactions within a colony affect these decisions? Answers to these questions are fundamental to our understanding of system organization, robustness and sustainability. In a recent study on ant colonies [3], we defined the stop-and-go signal through the instantaneous speed of individuals tracked over 100 min at 0.1 s inside the nest such that a ‘go’ is a sequence of time units characterized by a non-zero instantaneous speed bracketed, on either side, by a time unit with zero instantaneous speed [3]. We discovered a sublinear positive power-law relationship between the movement duration and the average instantaneous speed. The amount of social interaction had a significant effect on the exponent of this relationship because on average it was significantly greater when a colony resided in the larger of two experimental nests, both within the range preferred by this species, and hence at a lower density. This means that at a lower density the average instantaneous speed increases more quickly with increasing movement duration [3]. The relationship between movement duration and average speed is therefore flexible to environmental conditions. Yet it is also universal. When the speed for each time point was rescaled as a proportion of the average speed over the whole movement event and each time point was rescaled as a proportion of the total movement duration, the relationship collapsed onto the same profile for ants from each of the three studied colonies for each of the two nest sizes. This means the profile was the same irrespective of the nest size and hence the number of interactions and the exponent of the relationship between movement duration and average speed. This profile shows that on average ants move at the average instantaneous speed for a movement duration except for a short acceleration and deceleration at the beginning and end. The universal relationship between movement duration and average speed does not necessitate direct social interaction [4] even though, as stated above, it does affect its parameters [3]. This is because it holds not only for ants inside the nest [3] but also for the movements of lone scouts and foragers in large arenas outside the social home [4]. The power exponents are smaller and the constants are greater than for ants moving within the nest but despite the higher speeds at which ants move outside the nest, the rescaled relationship collapses onto the same profile of movement at the average speed for most of the movement duration [4]. Intriguingly, the universal relationship between movement duration and average speed has now been reported also in humans for their 24 h circadian cycle of activity [5]. Since in both ants and humans, the profile of the rescaled instantaneous speed as a proportion of the average speed over the duration of a movement event decreases towards the end of the event, there is strong evidence to suggest that the duration of movement is determined prior to its commencement [3-5]. Therefore, the arrow of causation points in the direction from movement duration to the average instantaneous speed associated with it [3-5]. This means that ants make the decisions about the start, duration and ending of a movement event during the preceding stationary period [4]. The universality of individual movement and the likely role of stationary periods as a time for decision-making has important implications for our understanding of task allocation, division of labour, food distribution and other phenomena at the social level that involve communication, coordination and synchronization of individuals. The power-law relationship between movement duration and average speed suggests that ant colonies can adapt quickly to different environments. Therefore, such a movement characteristic might facilitate an efficient information transfer. To the best of our knowledge, intermittent locomotion has not yet been implemented in swarm robotics. ‘Resting’ has been considered for robots in the context of foraging as a behavioural state that is more likely when the individual is not successful at finding food [6–9]. As the name suggests, the function of this state is to save energy. My aim is to highlight the importance of (1) behavioural intermittency, (2) the universal but flexible relationship between movement event characteristics and (3) the very likely decision-making function of the concomitant stationary periods for the sustainability and robustness of collective systems. Future understanding will be facilitated by a feedback between biological experiments and theory as well as by results from virtual and embodied swarms of robots.

14:50-15:50 Session WeP1c: Bio-inspired studies (3)
Location: Room I
14:50
Intelligent Material Handling of Mobile Robots with an Uncalibrated Binocular Vision System

ABSTRACT. An autonomous mobile robot was constructed in this work. The developed mobile robot is mainly composed of a mobile base, a robot manipulator, and a binocular vision system in eye-to-hand configuration. This flexible material transfer system can move materials from one station to another. During pick-and-place operations between a predefined station and the mobile robot, position and orientation errors of the mobile base are inevitably caused by the guidance control system of the mobile base. To compensate for these errors, a vision-guided control strategy is proposed to drive the manipulator to reach the specified location in the station to perform a pick or place operation. The proposed control strategy is based on the extended Kohonen’s network of the self-organizing map, which is applied to achieve the nonlinear mapping relationship between image data and the position of the manipulator after some learning steps. With the characteristic of self-learning, the calibration for the whole robotic system is unnecessary. Finally, the positioning performance of the manipulator is experimentally evaluated for the material in various locations on a station after the learning phase and the proposed control strategy is illustrated by a pick operation.

15:10
Dynamical Analysis of Pace and Trot Gait Using a 3D Quadrupedal Model

ABSTRACT. Quadrupeds selectively use pace and trot gait. However, the reasons why animals do pace or trot depending on their species or situations are still unclear. If we understand dynamical differences of them, it should lead to reveal this problem. In this paper, we use a simple 3D quadruped model to find periodic motions corresponding to pace and trot, respectively. Then we compared obtained motions from the viewpoints of energy saving and the maximum ground reaction force. From the results, we found that there is a trade-off relationship between two gaits.

15:30
Investigation of Quadrupedal Bounding Gait Using a Linearized Model with the Body Flexibility

ABSTRACT. The observation of quadruped animals has suggested that the body flexibility is important for fast running. In particular, the body flexibility is assumed to be crucial for determining the ground reaction force (GRF) on legs. In this paper, we investigated the dynamic effect of the body flexibility on GRF in quadruped bounding with simple physical model by focusing on the vertical motion. Furthermore, to clarify the GRF reduction mechanism, we made a linearized model from the two legged model with flexible body. We investigated the varidity of the linearized model by comparing the impulse on legs, instead of GRF, with the original two legged model and show that the use of the body flexibility reduces the impulse also in the linearized model.

14:50-15:50 Session WeP2c: Swarm science (3)
Location: Room II
14:50
The Decision Making of Admission to Membership and Participation in a Group Influences the Evolution of Group Cooperation

ABSTRACT. Cooperation in groups is the foundation of not only animals and microbes but also humans. However, the evolution of cooperation in a group is an evolutionary puzzle because defectors always obtain a higher benefit than cooperators. When people participate in a group, they evaluate the group members and then decide whether to participate in it. In some groups, membership is open to all those who are willing to participate in the group. In other groups, a candidate is admitted to membership if group members regard the candidate’s reputation as good. We developed an evolutionary game model and investigated how participation in groups and admission to membership influence the evolution of cooperation in groups where the public goods game is played by means of agent-based simulation. When the group's membership is open to all candidates and candidates can decide whether to participate in a group, cooperation cannot be sustainable. However, cooperation is sustainable when a candidate can be a member if all members in the group admit them to membership. Therefore, it is not participation in a group but rather admission that is essential to sustain group cooperation.

15:10
Building Cellular Artificial Agents to Interface with Cells in Living Tissues

ABSTRACT. The form and function of biological tissue derive from collective interactions among the thousands to millions of cells within the tissue. Such cell-cell interactions are crucial for processes in embryonic develop, cancer metastasis, and wound healing. As with any collective system, these interactions rely on information coupling across the individual agents. Hence, just as a robot coated in cockroach pheromones can influence cockroach behavior, a synthetic object can be coated with unique cellular proteins in order to trick cells into recognizing the object as ‘cell-like’. In this work, we demonstrate how this can be accomplish by coating 3D microstructures with purified proteins that uniquely mimic cell-cell recognition and adhesion, and we show how these new cell-mimetic materials can be used to infiltrate into, and integrate with living tissues in order to control their behavior from the inside-out.

15:30
Group-size Regulation in Self-Organised Aggregation through the Naming Game

ABSTRACT. Language games are self-organised models of the evolution of language used in the context of natural language evolution. They are used to study the emergence of a shared vocabulary through a self-organisation process. These models have been traditionally tested in settings where the interaction topology among individuals is mostly static. Only recently, these models have been introduced to the context of swarm robotics, in order to study the effect of embodiment and random agent mobility on the evolution of language. These results have shown that, even in such setting, the naming game exhibits the same outcome as in simpler simulations: all agents achieve consensus on a single word. In this paper, we study the interaction effect between the naming game and one of the simplest, yet most important collective behaviour studied in swarm robotics: self-organised aggregation. This collective behaviour can be seen as the building blocks for many others, as it is required in order to gather robots, unable to sense their global position, at a single location. Achieving this collective behaviour is particularly challenging, especially in environments without landmarks. Here, we augment a classical aggregation algorithm with a naming game model. Experiments reveal that this combination extends the capabilities of the naming game as well as of aggregation: It allows the emergence of more than one word, and allows aggregation to form a controllable number of groups. These results are very promising in the context of collective exploration, as it allows robots to divide the environment in different portions and at the same time give a name to each portion, which can be used for more advanced subsequent collective behaviours.

16:00-17:00 Session WeP1d: Bio-inspired studies (4)
Location: Room I
16:00
Relationship between robustness of ant colonies and behaviour change based on global information

ABSTRACT. The swarm is an autonomous decentralized system that performs complicated tasks despite having no colony’s global information. We consider that each individual changes its behaviors based on estimated global information. In this paper, we consider a red harvester ant colony. An ordinary differential equation (ODE) model of red harvester ants’ behaviour is constructed. Simulations are carried out to demonstrate our conjecture.

16:20
Three psychophysical laws in honeybee house-hunting

ABSTRACT. Psychophysics studies the relationship between stimulus intensity and its perception in the human brain. This relationship has been explained through a set of psychophysical laws that hold on a wide spectrum of sensorial domains. More recently, numerous studies have shown that a wide range of organisms at various levels of complexity obey these laws. In this study, for the first time, we show that superorganismal behaviour, such as honeybee nest-site selection, may obey the same psychophysical laws displayed by humans in sensorial discriminatory tasks. In our previous work, through stability analysis of deterministic models, we evidenced colony-level regime changes that could possibly lead to dynamics in agreement with Weber's law. In this study, through a thorough analysis and stochastic computational simulations, we investigate the adherence to three psychophysical laws: Weber's, Hick-Hyman's, and Pieron's law. Finding psychophysical laws in a non-neurological model contributes to the identification of general mechanisms generating these patterns and the adaptive benefits that led to them. Our results confirm that a superorganism can display cognitive abilities with characteristics comparable to higher order organisms.

16:40
Phototactic Supersmarticles

ABSTRACT. Smarticles, or smart particles, are small robots equipped with only basic movement and sensing abilities that are incapable of rotating or displacing individually. We study the ensemble behavior of smarticles, i.e., the behaviors a collective of these very simple computational elements can achieve, and how such behaviors can be implemented using minimal programming. In this work, we show that an ensemble of smarticles, constrained to remain close to one another (which we call a supersmarticle), achieves directed locomotion towards or away from a light source, a phenomenon known as phototaxing. We present both experimental and theoretical models of phototactic supersmarticles that collectively move with a directed displacement in response to light. The motion of the supersmarticle is Brownian, and is a result of chaotic interactions between smarticles. The system can be directed by introducing asymmetries in certain individual smarticle's behavior, in our case by varying the smarticles' activity levels in response to light, resulting in macroscopic supersmarticle biased motion.

16:00-17:00 Session WeP2d: Swarm science (4)
Location: Room II
16:00
Ecological and environmental factors limiting and inducing nest relocation in an ant species

ABSTRACT. We tracked migrations of colonies of Diacamma sp. from Japan in the field using a special whole-colony-level-mark-recapture method. We found a negative correlation between the broods/workers ratio and the migration distance. On average nests moved to the direction avoiding the nearest neighbor nests. Nests were uniformly distributed throughout the active season. Meteorological data suggested that rain fall and dry condition were likely the proximate stimuli triggering nest relocation. Feeding experiments suggested that food shortage was not an important factor for nest relocation. Those results suggest that the energetic cost of nest relocation and the competition between neighboring colonies are the factors limiting the nest relocation direction and the distance in this ant.

16:20
Reconsideration of Fixed Response Threshold Model and Its Application to Foraging Behavior of Ants

ABSTRACT. Task organization is one of the most interesting phenomena seen in colonies of social insects. Even for a single task, we can observe many characteristic aspects for the task organization, such as i) the hierarchy of the activity level among workers and ii) the response to the change of surrounding condition. To simultaneously understand the mechanism of both phenomena, we construct a model that focuses on foraging behavior based on the fixed response threshold model and examine the model through numerical simulations and analytical approaches.

16:40
Analysis of Dynamical Behaviour of Flying Pigeons Using INS/GPS Logger

ABSTRACT. Dynamical behaviour of flying pigeon is analyzed with data obtained by inertial sensors attached to them. Time series of acceleration and angular velocity are statistically compared to each other, and the flapping motion is quantified by autocorrelation functions.