ASSISIbf
Animal‌ and‌ robot‌ Societies‌ ‌Self-organise‌
and‌ Integrate‌ by‌ Social‌ Interaction‌ (bees‌ and‌ fish)‌

Modeling collective behavior: from physical particles to perceptual agents

For group living animals, collective behaviors often depend on perception of environmental cues and social interactions among group members. As large animal societies often lack a global communication system, one of the most challenging problems faced by these groups is probably the coordination of all group members. It requires gathering the information about the environmental opportunities, information transfer between group members and information processing by individuals. In large groups, individuals mostly respond to local information since they only have access to limited knowledge and are unable to directly compare the different environmental opportunities. Thus, collective pattern displayed by such groups frequently rely on decentralized processes based on amplifying loops that are base on direct interactions between individuals or through intermediate signals. Such mechanisms have been shown to rule various collective activities in numerous species, including humans (Figure 1, Bonabeau et al., 1997, Camazine et al., 2001, Couzin and Krause, 2003, Sumpter, 2006 and Moussaïd et al., 2009).

Figure 1. Examples of collective behaviors. (a) Fire ants forming a raft on water. (b) Honeybee swarm. (c) Starling murmuration. (d) Gnu migration. (e) Sardine school. (f) Human crowd.

Figure 1. Examples of collective behaviors. (a) Fire ants forming a raft on water. (b) Honeybee swarm. (c) Starling murmuration. (d) Gnu migration. (e) Sardine school. (f) Human crowd.

Among them, collective motion has been studied by researchers from numerous fields like biology, physics, computer science and robotics. It has raised an increasing interest these last decades and has led to a huge amount of experimental literature in fish schools, bird flocks, mammal herds or insect colonies. Facing this amount of data, scientists tried to infer the individual rules used by group members to produce the observed collective patterns. For this purpose, the increasing efficiency of the informatics and computers provide a very helpful tool by allowing the development of individual-based models.
The first models described group members interactions according to their spatial position: the movement of an individual is based on the position of a subset of its neighbors. In its pioneering work in computer animation inspired by the Particle Systems of Reeves (1983), Reynolds proposed in 1987 a sufficient set of rules to reproduce flocking behavior of birds by describing the motion of geometrical objects called boids, a contraction of bird-oids objects. These agents follow three simple rules: “attraction” –boids are attracted by other agents- “velocity matching” –boids align their velocity with nearby neighbors and “repulsion” –boids avoid collisions with too close neighbors (Figure 2).

Figure 2. Behavioral rules followed by the boids (in Reynolds, 1987; http://www.red3d.com/cwr/boids/). (a) Cohesion. (b). Velocity matching. (c) Repulsion.

Figure 2. Behavioral rules followed by the boids (in Reynolds, 1987; http://www.red3d.com/cwr/boids/). (a) Cohesion. (b). Velocity matching. (c) Repulsion.

The first demo of this model reproduces a flock of boids moving collectively and avoiding obstacles (http://www.siggraph.org/education/materials/HyperGraph/animation/art_life/video/3cr.mov) while it was firstly used for a computer animation movie in “Stanley and Stella in Breaking the ice” produce in 1987 (http://www.youtube.com/watch?v=3bTqWsVqyzE). A few years later, the same algorithm was used to produce bat swarms and penguins armies invading Gotham City in Tim Burton’s movie “Batman Returns” (Lebar Bajec and Heppner 2009).
In parallel to the development of computer-simulated flocks, the study of collective motion has also given rise to an increasing interest by physicists. In their wel1-known study of 1995, Vicsek and his collaborators focused particular attention on the emergence of self-ordered motion in systems of particles mimicking biological interactions. They described the motion of self-propelled particles interacting with other particles situated within a distance. At each time step, interactions forces between a particle and its neighbors are calculated and the particle update its position along the resulting force vector. In this model (and the numerous ones derived from it since almost 20 years), the main process is divided into 3 steps: i) gathering information, ii) processing information and iii) update its position.
Multiple hypotheses have been proposed to calculate the subset of individuals who influence the motion of a focal fish. As described above, the first ones were based on a metric perception: a fish takes information from all individuals situated within a distance. A second rule, called “topological perception”, states that fish move according to the position of their proximal neighbors. Another hypothesis based on Voronoi diagrams described fish movement as a function of a tessellation of their neighbors (Figure 3, for a comparison between these three postulates, see Strandburg-Pashkin et al., 2013). Although these models produce quite exhaustively the global patterns observed in flocking populations, we still lack clear experimental evidence to validate them.

Perception hypothesis

Figure 3. Possible types of perception. (a) Metric perception: all neighbors within a fixed distance. (b) Topological perception: the n proximal neighbors (5 in our example). (c) Voronoi perception: all individuals connected to the focal individual by a Delaunay triangulation.

For this reason, current experimental and theoretical research is now investigating the mechanism used by fish through a “sensory” perspective (Lemasson et al., 2009; Lemasson et al., 2013; Strandburg-Pashkin et al., 2013). In these models, the position of the neighbors of a fish are not described in Cartesian coordinates but are represented in the perception field of the focal fish and are thus projected on a simplified retina (Figure 4). Then, the model simulates the neurological processes to interpret the images received by the retina. Although the neurological processes remain mostly not understood, we are closer to an effective description of the individual behavior.

Figure 4. Example of visual perception. We simulated the motion of 20 agents in a finite space and computed the visual field of a focal individual (the green one). All agents are represented in the visual field according to their body length and orientation.

Figure 4. Example of visual perception. We simulated the motion of 20 agents in a finite space and computed the visual field of a focal individual (the green one). All agents are represented in the visual field according to their body length and orientation.

In parallel to the understanding of information processing by individuals, this approach is a new step towards bio-inspired algorithms that can be implemented in robotic agents. Indeed, a major scientific challenge, that is at the core of the ASSISIbf project, is to build artificial systems that can perceive, communicate to, interact with and adapt to animals (Schmickl et al., 2013). To do so, we need to develop artificial agents that communicate through appropriate channels corresponding to specific animal traits but also that correctly perceive and interpret signals emitted by the animals (Halloy et al., 2013, Mondada et al., 2013). This was firstly achieved in 2007 by building bio-inspired artificial cockroaches that where able to sense the presence of congeners and to adapt their behavior following a bio-inspired algorithm (Halloy et al., 2007). In this perspective and to continue in this path, the development of perception-based models is an obligatory step toward intelligent artificial systems capable of closing the loop of interaction between animals and robots.

References

Bonabeau, E., Theraulaz, G., Deneubourg, J.L., Aron, S., Camazine, S., 1997. Self-organization in social insects. Trends Ecol. Evol. 12, 188-193.

Camazine, S., Deneubourg, J.L., Franks, N.R., Sneyd, J., Theraulaz, G., Bonabeau, E., 2001. Self-organization in biological systems, Princeton University Press.

Couzin, I.D., Krause, J., 2003. Self-organization and collective behavior in vertebrates. Adv. Stud. Behav. 32, 1-75.

Halloy, J., Mondada, F., Kernbach, S. and Schmickl, T. 2013. Towards bio-hybrid systems made of social animals and robots. In Biomimetic and Biohybrid systems, Second International Conference, Living Machines 2013, Eds. Lepora, N.F., Mura, A., Krapp, H.G., Verschure, F.M.J. and Prescott, T.J., London, UK, Proceedings.

Halloy, J., Sempo, G., Caprari, G., Rivault, C., Asadpour, M., Tâche, F., Saïd, I., Durier, V., Canonge, S., Amé, J.M., Detrain, C., Correll, N., Martinoli, A., Mondada, F., Siegwart, R. and Deneubourg, J.L., 2007. Social integration of robots into groups of cockroaches to control self-organized choices. Science, 318, 1155-1158.

Lebar Bajec, I. and Heppner, F.H. 2009. Organized flight in birds. Anim. Behav., 78, 777-789.

Lemasson, B.H., Anderson, J.J. and Goodwin, R.A., 2009. Collective motion in animal groups from a neurobiological perspective: The adaptive benefits of dynamic sensory loads and selective attention. J. Theor. Biol., 261, 501-510.

Lemasson, B.H., Anderson, J.J. and Goodwin, R.A., 2013. Motion-guided attention promotes adaptive communications during social navigation. Proc. R. Soc., 280, 20130304.

Mondada, F., Halloy, J., Martinoli, A., Correll, N., Gribovskiy, A. Sempo, G., Siegwart, R. and Deneubourg, J.L. 2013. A general methodology for the control of mixed natural-artificial societies. In Handbook of collective robotics, Ed. Kerbach, S., Pan Stanford Publishing.

Moussaïd, M., Helbing, D., Garnier, S., Johansson, A., Combe, M., Theraulaz, G., 2009. Experimental study of the behavioural mechanisms underlying self-organization in human crowds, Proc. R. Soc. B, 276, 2755–2762.

Reeves, W.T. 1983. Particle systems – A technique for modeling a class of fuzzy objects. Comp. Graph., 17, 359-375.

Reynolds, C. 1987. Flocks, herds, and schools: a distributed behavioural model. Comp. Graph, 21, 25-34.

Schmickl, T., Bogdan, S., Correia, L., Kernbach., S., Mondada, F., Bodi, M., Gribovskiy, A., Hahshold, S., Miklic, D., Szopek, M. Thenius, R. and Halloy, J. 2013. ASSISI: Mixing animals with robots in a hybrid society. In Biomimetic and Biohybrid systems, Second International Conference, Living Machines 2013, Eds. Lepora, N.F., Mura, A., Krapp, H.G., Verschure, F.M.J. and Prescott, T.J., London, UK, Proceedings.

Strandburg-Pashkin, A., Twomey, C.R., Bode, N.W.F., Kao, A.B., Katz, Y., Ioannou, C.C., Rosenthal, S.B., Torney, C.J., Wu, H.S., Levin, S.A. and Couzin, I.D., 2013. Visual sensory networks and effective information transfer in animal groups. Curr. Biol., 23, R709-R711.

Sumpter, D.J.T., 2006. The principles of collective animal behavior. Philos. T. Roy. Soc. B, 361, 5-22.

Vicsek, T., Czirok, A., Ben-Jacob, E., Cohen, I. and Shochet, O., 1995. Novel type of phase transition in a system of self-driven particles. Phys. Rev. Lett., 75, 1226-1229.