Highlight
Autonomous Neuro-Evolving Robotic Devices (NERDs)
Robots are coordinated in their behavior around a common target
Achievement/Results
Graduate student Joshua de Leeuw from the IGERT program in brain-body-environment systems at Indiana University and colleagues at Vassar College have demonstrated that simple robotic fish can coordinate their behavior without any social communication. This work has potential applications in developing methods for controlling and coordinating the behavior of groups of robots, as it suggests a very simple and low-cost principle that can organize group behavior into stable patterns. It also furthers our understanding of coordinated behavior in groups of animals like flocks of birds, schools of fish, and swarms of ants.
The work involved extensive interdisciplinary collaboration. Joshua is a graduate student in cognitive science and psychology, and the research team at Vassar College included biologist and expert in biorobotics John Long, as well as biologist Marianne Porter, and cognitive scientist Ken Livingston. The research team designed and built a set of small, autonomous, swimming robots which could be programmed to behave randomly or swim towards a target. The robots were specifically designed to be unable to detect other robots in the tank. The researchers filmed groups of different sizes interacting in a small pool over a period of five minutes. They found that when the robots were programmed to swim towards the target, the robots not only gathered at the target but also became highly coordinated after some initial disorderly behavior. Despite not being programmed to do so and lacking any social communication, the robots exhibited a stable pattern of coordinated behavior.
The researchers demonstrated that the schooling was a result of all of the robots having a similar target and the physical dynamics of the robot’s swimming behavior. As part of their effort to analyze the behavior of the robots, the researchers improved upon existing analytical tools for quantifying coordinated behavior of a group. The IGERT program in brain-body-environment systems at Indiana University is specifically designed to foster the investigation of these kinds of emergent interactions between agents interacting with the environment. IGERT students in the program receive training in the principles of emergent dynamics, which were applied to this research project.
Address Goals
This activity addresses NSF’s strategic goal of Discovery by furthering our understanding of coordinated behavior in groups of animals such as flocks of birds, schools of fish, and swarms of ants. It also has potential applications to the coordination and control of groups of robots by suggesting a simple and low-cost principle that can organize group behavior into stable patterns. This activity addresses NSF’s strategic goal of Learning by engaging a Ph.D. student in research that crosses institutions, disciplines and methodologies.
