Nicole Abaid earns NSF CAREER award to study bio-inspired active sensing in multi-agent systems
Nicole Abaid, assistant professor in the Department of Biomedical Engineering and Mechanics in the College of Engineering, has received a National Science Foundation Faculty Early Career Development (CAREER) award to examine active sensing in natural systems and develop a dynamics and control framework for multi-agent sensory communication.
During the five-year grant, Abaid will collect data from wild gray bat swarms to study their collective behavior and interaction patterns, especially in regards to how these animals use echolocation. She will use these data sets to build and test a model for control in multi-agent robotic systems, deploying the model in an engineering problem known as simultaneous localization and mapping, or SLAM.
Bats are unique not only in the use of echolocation, or bio sonar, to navigate their environment, but also because of their ability to intercept these sensing signals from their peers. Previous research has shown that small groups of bats might change their behavior to avoid negative interference from such interceptions and they sometimes could even “eavesdrop” on others’ signals to navigate and hunt more efficiently.
“In engineering systems – like a team of robots, for example – active sensing happens when a robot generates a signal that interacts with its environment, and then the robot listens,” said Abaid. “But the fact that a signal could be intercepted by another robot is almost only ever a bad thing. It’s called jamming, and it’s like hearing extra noise.”
“But with bats, they don’t seem to suffer the ill effects that engineering systems do,” she said.
Using a custom-built setup of thermal cameras and microphone arrays, Abaid will observe bat swarms in the wild in collaboration with Mark Ford in the Department of Fish and Wildlife Conservation and the U.S. Geological Survey, as well as the Virginia Cooperative Fish and Wildlife Research Unit. She’ll then develop a mathematical model of their active sensing system and look for emergent behaviors in that system, translating her findings into a control algorithm for a multi-agent team of robots.
“We want to develop an algorithm that can do a better job of solving this really typical engineering problem, SLAM,” said Abaid.
Commonly used in robotic mapping and navigation, SLAM is the computational problem of constructing or updating a map of an unknown environment while simultaneously keeping track of a robot’s location within that map. SLAM is often used during robotic exploration of unknown or unsafe environments, and its applications include mapping the ocean floor as well as collapsed buildings after natural disasters. Even robot vacuums use a form of SLAM to map the rooms they clean.
In multi-agent SLAM systems, a team of robots is deployed instead of a single robot. Multi-agent systems may be more robust in unpredictable environments, but their current methods of communication are somewhat limited. For example, robot teams can’t effectively intercept each other’s sensing signals without suffering the effects of jamming.
“We’ll use these bat-inspired algorithms to couple sensing and communication in multi-agent robotic teams so each robot could build a partial map and then share that map with the others,” said Abaid. “That’s a hard problem, but we think this multi-agent solution may be more efficient than a typical single-agent solution and even current multi-agent ones.”
As part of the program, Abaid will collaborate with Michael Rosenzweig in the Department of Biological Sciences to build swarm-themed educational modules for use in the Virginia Tech Biological Sciences Outreach Program at the SEEDS – Blacksburg Nature Center, which supports the growth of K-12 participation in scientific research. These educational kits will be made available to teachers throughout southwest Virginia and will also be used in SEEDS summer camps.
Written by Emily Roediger