CoBe : Collective Behavior

Collective behavior is often studied across various scales and environments, using both mathematical models with subsequent agent-based simulations and observations of biological agents.

Motivated by synthetic collective intelligence, these models serve as a foundation for designing artificial agents, such as robots. These agents are used to test the model's robustness in the physical world and to achieve practical applications.

Throughout this workflow, numerous environmental gaps emerge that must be bridged: from observations of animals in the physical world to simulated agents, and then back to the physical world with robotic agents. CoBe is a platform designed to bridge these gaps by providing a unified environment for humans, simulated agents, and robots to explore the role of embodiment in collective behavior.

Next

CoBe : AR for humans

CoBe is an augmented reality (AR) setup designed by David Mezey, Dr. Palina Bartashevich and David James with support of the Science of Intelligence Cluster. It is a platform to connect simulated and physical worlds.

CoBe is a set of projectors and cameras spatially distributed on a scaffold around an active arena all connected through a controller station. CoBe detects dedicated objects, adjusts the underlying simulation environment based on their positions, and projects these changes back into the arena, creating a dynamic, interactive space. CoBe can be flexibly modified to detect any object and run any real-time simulation defining the environment to be projected.

In its default mode CoBe runs a computational model by Dr. Palina Bartashevich exploring predator-prey interactions in simulated groups of fish. CoBe projects a school of virtual fish that move within the arena. Humans can interact with the group of prey via a special staff representing a predator for the virtual fish, to which they react with collective escape behavior also exhibited by real fish. Thus one or multiple humans can act as predator(s) in the system and try to control the fish school.

Next

CoBe : AR for robots

CoBe also supports Thymio II robots providing a unique tool in the ongoing and future projects of David Mezey. In this case robots can sense the projected fish swarm either via their cameras (decentralized) or via a central information node (centralized position coordinates). The robots then can behave in the physical world interacting with the simulated fish.

By integrating robots in CoBe we have the unique possibility to bridge the gap between simulated and physical agents, and among others, (1) to study the differences between centralized and decentralized control, (2) to quantify the effect of perceptual noise on agent behavior or (3) to explore the effect of physical constraints such as movement limitations coming from the embodiment of agents.

Furthermore, CoBe can also simplify experimentation with robot collectives. In our case, from the 2 different types of agents (predators and prey) we only need to physically engineer one of them keeping the problem tractable.

Next

CoBe : AR stands for art

CoBe also serves as a tool for scientific outreach and artistic visualization. The rendering of simulated fish was inspired by observations from the Gulf of Mexico, where the interactions between blue marlins and mackerels also inspired the underlying mathematical model.

As a tool for scientific communication, CoBe incorporates multiple artistic and user-facing elements. From rendering the model into visuals, to projection mapping, to enhanced believability and even animating the fish by hand, David James' efforts bring to life an otherwise unobserved dimension of the system.

The background music, specifically composed for CoBe by Mui Zyu and Natalie Szende, enhances the audiovisual experience. This collaboration aims to create an immersive environment that bridges science and art, inviting deeper engagement with the simulated ecosystem.

Next

Videos

CoBe : Humans

A pair of humans chasing "fish"

CoBe : Robots

A pair of robots chasing "fish"

CoBe : Art

Our credits video with original footages from the Gulf of Mexico