Bacteria, bees, and birds often work together in groups to find food. A group of mobile wheeled robots can be designed to coordinate their activities to achieve a goal. Networked cooperative autonomous air vehicles are being developed for commercial and military applications. In order for such multiagent systems to succeed it is often critical that they can both maintain cohesive behaviors and appropriately respond to environmental stimuli. In this talk, we characterize cohesiveness of discrete-time multiagent systems as a boundedness or stability property of the agents' position trajectories and use a Lyapunov approach to develop conditions under which local agent actions will lead to cohesive group behaviors even in the presence of (i) an interagent "sensing topology'' that constrains information flow, where by "information flow,'' we mean the sensing of positions and velocities of agents, (ii) a random but bounded delay and "noise'' in sensing other agents' positions and velocities, and (iii) noise in sensing a resource profile that represents an environmental stimulus and quantifies the goal of the multiagent system. Simulations are used to illustrate the ideas for multiagent systems and to make connections to synchronization of coupled oscillators.
Division of labor, the way in which social groups distribute work among their individual members, is a product of self organization and selection. A basic system of division of labor can be produced even in artificial associations of normally solitary individuals and fits simple rules of interaction. In social insect colonies, however, the process of division of labor reflects the integration of the colony itself. I will discuss how division of labor changes with increased group size within social insect (primarily ant) colonies and discuss a network subgraph approach for capturing colony integration and regulation within social insect colonies s.
As social insects have evolved division of labor and colony organization to accomplish tasks necessary to their survival, their social and collaborative environment should make them more and more susceptible to risk from infectious disease. Since they haven't been forced to extinction yet, they're clearly doing something right. Some have evolved individual physiological protections, others have behaviorally mediated individual responses/defenses, and a few have been shown to have collaborative behavioral defenses. In this talk, we'll discuss a set of models that explore whether or not the entire social organization of colonies themselves shows evidence of evolutionary selective pressures from disesase risks.
The ant society is a dynamic network of interacting nestmates of which individual decision rules lead to adaptive and functional patterns at the collective level. The non-linearity of relationships between workers makes those societies displaying properties characterizing other complex systems such as a high sensitivity to the number and/or rates of interactions between system agents. In the case of ant foraging, exploitation patterns strongly depend on colony size in a non-linear and discontinuous way: the density of nestmates' interactions influences the occurrence as well as the transition from one system state to another. However, ants do not passively undergo such a density-dependent structuring effect but instead, can play an active role in tuning feed-backs loops as a function of the density of workers. We shall review the ways workers can "assess" nestmates' density through either direct or indirect cues and then can tune amplification processes such as the laying of trail recruitment according to the social context of food exploitation. Another feature of ant societies as complex dynamic systems is the occurrence of hysteresis in which prior states of ant densities influence the ways through which the whole system can evolve. We shall see how ant individuals could keep track of such prior states and accordingly tune their behaviour and communication to improve their foraging efficiency. Finally, we shall discuss how the number of foragers can also deeply influence collective choices of ant societies between resources of different values and may act in conjunction with the availabilities of food resources of poor quality upon the discriminative abilities of insect societies
The amazing abilities of social insects to solve their everyday-life problems, also known as swarm intelligence, have received a considerable attention the past twenty years. Among their collective behaviors, nest building is certainly the most spectacular. Not only the characteristic scale of the nests is typically much larger than the size of the individuals, but some of the architectures built by insect colonies can also be highly complex. All along the evolution of these animals, there has been a whole set of innovations in terms of architectural designs and construction techniques that proved to be efficient to solve a large number of problems such as controlling the nest temperature, ensuring the gas exchanges with the outside environment or adapting the nest structure to various colony sizes. One fundamental question is: how large-scale patterns are generated by the actions and interactions of individual insects? To investigate this issue, we focused on the early stages of nest construction in the ant Lasius niger. This experimental paradigm was used to disentangle the coordinating mechanisms at work and characterize the individual behaviors (transport and assemblage of construction material). We then developed a 3D model implementing the mechanisms detected on the individual level and showed that they correctly explain the construction dynamics and the patterns observed at the collective level for various conditions. The model also revealed that complex helicoidal structures connecting nearby chambers emerge from a constant remodeling process of the nest architecture.
Twenty years ago, the case for optimality theory in evolutionary biology was set out in a review by Geoff Parker and John Maynard Smith. Thinking of what idealised animals should do if they are behaving optimally has informed behavioural ecology since its inception. With some exceptions, the study and theory of collective behaviour seems to be much more more mechanistic. This is probably because the mechanisms of collective behaviour are much more easily observed than those underlying individual decision-making, and because simple mathematical models and computational simulation often give good descriptions of collective systems. I shall argue that optimality theory is important for collective behaviour, review existing and potential applications of it, and highlight the crucial importance of selecting the right optimality criteria for a particular system.
Many animals take part in flow-like collective movements. In most species, however, the flow is unidirectional. Ants are one of the rare group of organisms in which flow-like movements are predominantly bidirectional. This adds to the difficulty of the task of maintaining a smooth, efficient movement. Yet, ants seem to fare well at this task. Do they really? And if so, how do such simple organisms succeed in maintaining a smooth traffic flow, when even humans experience trouble with this task? The experimental study of ant traffic is only a few years old but it has already provided interesting insights into traffic organization and regulation in animals, showing in particular that an ant colony as a whole can be considered as a typical self-organized adaptive and highly flexible system.
Agent-based dynamical models have been used successfully to reproduce a range of observed collective behaviors in biological groups. In these models, agents interact with one another and it has been shown that the topology of the interaction network plays a significant role in emergent outcomes and performance at the level of the group. An important challenge is to understand the tradeoffs, sensitivity to parameters, and different regimes of behavior in these biological models from the perspective of evolution by natural selection. Here we focus our attention on collective migration, defined broadly to represent a class of problems in which individuals in a group respond to an environmental cue and to social interactions. Models of collective migration have shown that a small group of leaders (individuals who invest strongly in the environmental cue) is capable of guiding a larger group of followers (individuals that rely on social interactions). Further, evolutionary simulations of migration models have shown that the speciation of a homogeneous group into leaders and followers is a stable evolutionary outcome when the cost of leadership is sufficiently high. Analytical mean-field evolutionary models using the techniques of adaptive dynamics have confirmed the observations in these simulations. We study the role that the interaction topology plays in the evolutionary outcomes of collective migration. As a point of comparison, we show that our model recovers the (qualitative) results of the mean-field analysis in the limit of all-to-all interconnections. We then demonstrate a minimum connectivity threshold for random interconnection graphs to yield speciated outcomes. We also study the adaptation of nodes on fixed graphs and illustrate the influence of graph topology on emergent outcomes in such adaptive systems.
The application of self-organization theory to social insect studies is, for the most part, barely 20 years old. It has been remarkably successful because much of the new thinking and modelling that self-organization theory has brought to social insect studies has been very provocative, sometimes naive, and often oversimplifying; yet it has, almost invariably, lead to new experiments that have formed foundations for further progress. This has been a tale not of vicious circles but of virtuous ones. They are virtuous because errors and misunderstandings are exposed and corrected. They have gained great momentum from the natural, yet uneasy, tension between mathematical and empirical explanations. But most of all, they have been successful because mathematical modellers and experimentalists have worked together intimately both on the models and the experiments. In this talk, my aim is to illustrate these principles and the success of this endeavour by reviewing certain key examples. My goal is for this celebration of science past to suggest some of what might lie ahead.
Biological systems, from embryos to social insects, get tremendous mileage by having vast numbers of cheap and unreliable individuals cooperate to achieve complex goals. We are also rapidly building new kinds of distributed systems with similar characteristics, from multi-modular robots and robot swarms, to vast sensor networks. Can we engineer collective systems to achieve the kind of complexity and self-repair that nature seems to achieve?
In this talk, I will describe several projects from my group where we have used inspiration from nature -- termites, fireflies, and cells -- to design new kinds of robots and networks. For example, simple robots that collectively build structures without explicit communication, self-adaptive modular robots that respond to the environment, and wireless sensor networks that use firefly-inspired algorithms to achieve high throughput. In each case, we use inspiration from biology to design simple decentralized cooperation, and techniques from computer science to analyze and generalize these algorithms to new tasks. A common theme in all of our work is understanding self-organizing multi-agent systems: how does robust collective behavior arise from many locally interacting agents, and how can we systematically program simple agents to achieve the global behaviors we want.
In human children, parallel play describes two or more children playing side by side, perhaps using the same toy but for different purposes, and only occasionally modifying their behavior in response to the other. It forms an early stage of social development, following solitary play and generally preceding social and cooperative play.
If a group of ants were overseen by an extremely scientific teacher, how would he or she classify their interactions? I will address this question by studying models of three long-term interactions within and between ant colonies.
1. Ants must allocate effort among tasks such as foraging in different spatial locations, and do so based on information about what others, including nearby competitors, are doing.
2. Ants may need to choose conflict strategies to deal with neighbors of different species with different behaviors and fighting abilities, but without prior knowledge of who they will encounter.
3. Ants must choose strategies to compete with nearby or distant neighbors, potentially acting more or less aggressive toward members of familiar colonies.
Individuals can only base decisions on what they know, whether shaped by personal experience or shared information, ideally contributing to the long-term success of their colony. I will examine how well ants can regulate foraging and conflict with only limited information, and discuss when the resulting behaviors can be considered a coordinated strategy by the colony rather than "parallel work" by socially unsophisticated individuals.
In this talk, I will first highlight the challenges related to the design, control, modeling, performance evaluation, and optimization of distributed, mobile, resource-constrained robotic systems. In particular, I will describe a specific distributed control method based on multiple modeling levels which has provided up to date interesting results in several case studies concerned with distributed sensing and manipulation missions, investigated either by us or other research groups worldwide. I will support the discussion with a few concrete examples concerned with aggregation and assembling tasks. Finally, I will revisit our engineering methodology and outline its similarities, differences, and possible links with the world of social insects.
I will summarize some work on the link between individual behaviour and the dynamics of the swarm/flock. I will highlight two projects:
1. The behaviour of a 2D flock of aquatic birds, and how Ryan Lukeman (former PhD student, now at St FX University) figured out the underlying individual rules
2. models for social foraging, an ongoing project in my group joint with Nessy Tania, Ben Vanderlei and Joel Heath.
No description Available.
Locust plagues are one of the most infamous insect scourges, invading vast areas of Africa, Asia, Australia and the Americas. The reason that locusts form plagues is that they have an extraordinary capacity to change from shy, green, harmless grasshoppers into brightly coloured, swarming creatures when they experience crowding. This remarkable change can occur within the life of a single animal: the genome of the insect codes for both forms. I show that an important trigger for the change is bumping into other locusts. Stimulating touch-sensitive hairs on the back legs causes a rapid shift in behaviour, such that locusts become attracted to each other, rather than avoiding one another. Having identified the source of sensory stimulation that induces behavioural gregarization, we next analysed the associated neurochemical pathways involved and have recently shown that a pulse of serotonin causes the shift in behaviour upon crowding. Once a local aggregation reaches a critical number of insects, the locusts suddenly start to move as one. Using self-propelled particles models from statistical physics we have shown that this decision to start migrating does not involve leader locusts, but rather emerges collectively as a result of local interactions between individuals. Continuing to move as a group involves something very sinister, however. To illustrate, I next turn to another swarming animal, the Mormon cricket of North America. The reason these animals form vast marching bands is because they are seeking protein. The most abundant source of protein in a swarm of crickets is other crickets. The reason why they keep marching is that, if an insect stops, it gets cannibalized by the crickets coming from behind: they are on a forced march for protein. The same is true for locusts. The search for protein turns out to be a powerful force in shaping the biology not only of crickets and locusts, but of all animals - including humans. We have shown using experiments based on state-space geometric models for nutrition that many animals have a powerful appetite for protein. I show in humans that this protein appetite plays a key role in obesity. Protein comprising a minor part of our total energy budget, yet its intake is strongly regulated. I show how this combination leads to protein having the power both to drive the development of obesity - and to assuage it. Finally, I consider why it should be that many animals, humans included, should possess specific mechanisms that prevent overconsumption of protein. Using geometric models of nutrition I show that there are costs to over-consuming protein, and that the prevailing view that caloric restriction prolongs life is wrong - in insects at least.
P. S. Krishnaprasad
Geometric methods in control theory have had a useful role in the investigation of dynamics of collectives. In this talk, we build on models from this theory to sketch recent progress in understanding small networks governed by interaction strategies associated with pursuit. We extend these ideas to a broader array of variational principles in networks of interacting systems. Using symmetry and reduction methods, hamiltonian structures, and conservation laws, we explore questions of collective behavior. These results also suggest how such principles may be exploited in the extraction of individual behaviors from movement data on flocks and swarms.
We study equilibrium configurations of swarming biological organisms subject to exogenous and pairwise endogenous forces. Beginning with a discrete dynamical model, we derive a variational description of the continuum population density. Equilibrium solutions are extrema of an energy functional, and satisfy a Fredholm integral equation. We find conditions for the extrema to be local minimizers, global minimizers, and minimizers with respect to infinitesimal Lagrangian displacements of mass. In one spatial dimension, for a variety of exogenous forces, endogenous forces, and domain configurations, we find exact analytical expressions for the equilibria. These agree closely with numerical simulations of the underlying discrete model. The exact solutions provide a sampling of the wide variety of equilibrium configurations possible within our general swarm modeling framework. The equilibria typically are compactly supported and may contain d-concentrations or jump discontinuities at the edge of the support. We apply our methods to a model of locust swarms, which are observed in nature to consist of a concentrated population on the ground separated from an airborne group. Our model can reproduce this configuration; quasi-two-dimensionality of the model plays a critical role.
Work done in collaboration with Chad M. Topaz.
No description available.
Emergent patterns of flocks and swarms are at once beautiful and mysterious. We ask ourselves: "How and why do individuals coordinate these complicated maneuvers?" More specifically: how does self organization at lower levels influence emergent properties at higher levels? I will present the results of some of my studies addressing these questions using whirligig beetles and computer models to understand the three-part transition from (1) individual behavior to (2) group position to (3) the emergent behavior of swarms. Whirligig beetles make an ideal organism for studying general grouping phenomenon, such as those found in birds and fish, because they are composed of unrelated individuals, unlike bees and ants where altruism and kin-selection complicate the interpretation of emergent properties.
Working from the bottom up, we have used computer models and ethograms to examine how differences in (1) individuals influence the lower-level movement rules of the beetles. For example, sex, hunger, and age may influence an animal's preferred distance from others, and the speed with which they swim. These rules then influence the (2) group position which they occupy. For example, depending on predators and water speed, certain classes of whirligigs reliably end up at the edge or front of groups. We have modeled and carried out experiments which show that these positions are consistent with the hypothesis of self-organization: simple movement rules can explain the observed within-group segregation. Also, we present evolutionary optimization models that support the hypothesis that these group positions are individually adaptive. Finally, the group as a whole exhibits measurable behaviors (3) that seem to be an emergent property of levels one and two. These emergent properties include: group speed, turning, stopping, and mass predator escape. In conclusion we will discuss if the behaviors at all three of these levels are evolutionarily adaptive, or whether some might be neutral byproducts of an adaptive response at a different level.
I will present a simple adaptive network model describing recent insect swarming experiments. By exploiting an analogy with human decision-making models and considering network-like interactions, this model captures the experimental dynamics using a low dimensional system of equations that permits analytical investigation. It reproduces several characteristic features of swarms, including: spontaneous symmetry breaking, noise- and density-driven order-disorder transitions that can be of first or second order, intermittency, and metastable configurations displaying memory effects. By considering only minimal components, and introducing few elements of the spatial dynamics, it highlights the essential elements required to reproduce the observed behavior.
I will review what is known about one of the most enchanting forms of collective animal behavior: the skillful choice of a new home by a swarm of honey bees. The challenge has been to understand how the 1.5 kilograms of bees in a swarm, like the 1.5 kilograms of neurons in a brain, are organized so that even though each individual has limited information and limited intelligence, the group as a whole makes first-rate collective decisions. I will describe how this complex phenomenon has been analyzed through a combination of empirical studies (observations and experiments) and mathematical studies (simulation models). In general, the empirical studies have revealed how the bees act and interact to produce the abilities of whole swarms, and the mathematical studies have clarified why the bees behave as they do to create a reliable decision-makng system.
Nest site selection and swarm guidance in swarms of Apis mellifera are well studied, both observationally and theoretically, but not nearly so much is known about decision-making behaviour in other species of Apis. The Asian red dwarf honey bee, Apis florea, is an open-nesting honey bee, found in Southeast Asia, India and parts of the Middle East whose nest is a single comb in the midst of a cluster of bees formed around a small, shaded branch. As in A. mellifera, scouts go out from an A. florea colony that is looking for a new home and seek out suitable nest sites. They then return and dance to indicate the location of suitable new nest sites, but their dances are more variable and less intense than those of Apis mellifera and several sites may still be being advertised when the swarm takes off. In A. mellifera, scouts, who are informed about the location of the new nest site guide the swarm to their destination by flying rapidly through the swarm in the direction that the swarm needs to travel. In A. florea it is possible that different groups of scouts are directing the swarm in different directions. Using both observations and models that have been particularly devised for flying bees which have constantly changing speed, we will examine the process of swarm guidance in A. florea and explore what happens in a migrating swarm when different groups of scouts direct the swarm to different nest sites.