Presented By: The Center for the Study of Complex Systems
Complex Systems Webinar | Designing and Analyzing Collective Behavior in Natural and Robotic Swarms
Justin Werfel, Harvard University Wyss Institute for Biologically Inspired Engineering
A COMPLEX SYSTEMS WEBINAR
BLUEJEANS MEETING: https://bluejeans.com/138645612
ABSTRACT: Nature is full of systems where independent, limited agents together give rise to sophisticated collective results, from cells in a body to organisms in a society. Such cases show that dynamically interchangeable, individually unreliable components can produce effective and reliable outcomes at the group level. The goal of engineering systems that work in such a way has the promise of producing powerful systems with advantages like robustness to failures of individual elements and flexibility in adapting to unknown conditions and unexpected events. At the same time, meeting this goal requires overcoming challenges not only of accommodating unpredictability and limitations of individual agents, but also of designing low-level behaviors that guarantee a particular high-level outcome.
In this talk, I will discuss both natural and artificial systems that perform collective construction, a task where independent agents jointly build large-scale structures. I will first present a robotic system in which a user can specify a precise target structure as an input, and independent climbing robots follow simple rules that provably guarantee the correct completion of that structure. Key principles that make engineering such a system possible include using the environment as a tool for coordinating effort and facilitating mechanical tasks, and creating regularities that constrain the space of possible trajectories. Next I will give an overview of our experimental work with mound-building termites, identifying cues that individual workers use to help direct their actions and coordinate colony activity. These studies point to the importance of factors including active excavation, surface geometry, and humidity, but, surprisingly, show no role for the putative cement pheromone that has been central to the traditional understanding for six decades. The robot and insect studies complement one another, each providing new tools and productive directions for ongoing work in the other.
BLUEJEANS MEETING: https://bluejeans.com/138645612
ABSTRACT: Nature is full of systems where independent, limited agents together give rise to sophisticated collective results, from cells in a body to organisms in a society. Such cases show that dynamically interchangeable, individually unreliable components can produce effective and reliable outcomes at the group level. The goal of engineering systems that work in such a way has the promise of producing powerful systems with advantages like robustness to failures of individual elements and flexibility in adapting to unknown conditions and unexpected events. At the same time, meeting this goal requires overcoming challenges not only of accommodating unpredictability and limitations of individual agents, but also of designing low-level behaviors that guarantee a particular high-level outcome.
In this talk, I will discuss both natural and artificial systems that perform collective construction, a task where independent agents jointly build large-scale structures. I will first present a robotic system in which a user can specify a precise target structure as an input, and independent climbing robots follow simple rules that provably guarantee the correct completion of that structure. Key principles that make engineering such a system possible include using the environment as a tool for coordinating effort and facilitating mechanical tasks, and creating regularities that constrain the space of possible trajectories. Next I will give an overview of our experimental work with mound-building termites, identifying cues that individual workers use to help direct their actions and coordinate colony activity. These studies point to the importance of factors including active excavation, surface geometry, and humidity, but, surprisingly, show no role for the putative cement pheromone that has been central to the traditional understanding for six decades. The robot and insect studies complement one another, each providing new tools and productive directions for ongoing work in the other.
Related Links
Explore Similar Events
-
Loading Similar Events...