Collective dynamics and model verification: Connecting kinetic modeling to data

Continuum Abstractions for Scalable Control of Robotic Swarms with Minimal Capabilities and Information

Spring Berman

Arizona State University


Novel low-cost autonomous robots such as micro-aerial vehicles are currently being developed as a result of recent advances in computing, sensing, actuation, power, control, and 3D printing technologies. Massive populations, or swarms, of such robots have the potential to collectively perform tasks over very large domains and time scales, succeeding even in the presence of failures, errors, and disturbances. In this talk, I present a top-down approach for reliably controlling robotic swarms to achieve desired collective behaviors using only local sensing and common broadcast information. This approach relies on ordinary and partial differential equation models of swarm population dynamics that describe the robots’ roles, task transitions, and motion. The framework is applied to tasks that require (a) target spatial distributions of robot activities (such as data collection) over a domain; (b) target robot allocations along specified curves and around the boundaries of regions or structures; and (c) cooperative manipulation of heavy payloads. I describe methods for designing the robot control policies to be robust to environmental variations and to mimic experimentally observed behaviors of ants performing collective transport. Potential applications include environmental monitoring and exploration, search-and-rescue, disaster recovery, security operations, automated construction and manufacturing, and even biomedical imaging and drug delivery at the nanoscale.