A new swarm AI project takes on safety at scale | Penn Today
Source: https://penntoday.upenn.edu/news/penn-engineering-xlab-new-swarm-ai-project-takes-safety-scale
Archived: 2026-04-23 17:13
A new swarm AI project takes on safety at scale | Penn Today
Skip to Content
Skip to Content
News from
University of Pennsylvania
Try Advanced Search
What happens when artificial intelligence agents leave the digital world and enter the physical world? Researchers at Penn’s
School of Engineering and Applied Science
are tackling that question through a new three-year, international collaboration on Swarm AI, led by
Rahul Mangharam
, professor in electrical and systems engineering and principal investigator of the
Safe Autonomous Systems Lab
(xLAB). The project brings together three universities to study how large teams of physical AI agents can cooperate, compete, and act safely.
View large image
Ph.D. student Hongrui Zheng (left), and Rahul Mangaharam, ESE professor and principal investigator of xLAB, with the drones used in this research.
(Image: Courtesy of Penn Engineering)
“Most of today’s AI agents live purely in software,” says Mangharam. “We’re moving toward physical AI, systems that don’t just generate answers, but act in the real world. And once AI operates in physical space, it has to deal with real constraints and real consequences.”
Unlike digital agents, physical agents must obey the laws of physics. They must avoid collisions, respect safety boundaries, and coordinate with teammates. The
research
centers on cooperation and coordination in adversarial games—scenarios in which teams of agents must strategize against opponents while maintaining internal cohesion.
“A key technical focus is understanding intent,” says Mangharam. “Agents must infer what other agents, human or machine, are trying to achieve and adjust accordingly. They must coordinate without centralized control and respond to dynamic, uncertain environments. This project brings together research in machine learning for multi-agent systems that use game theory to cooperate and compete.”
At large scales, this becomes an algorithmic and a systems problem: how to design distributed algorithms that scale to tens, hundreds, or even thousands of agents to make consistent, safe decisions in real time. One distinctive aspect of the project is its emphasis on neurosymbolic AI, which combines neural networks with structured, human-encoded knowledge.
“You can’t just throw AI at a problem and expect it to magically figure everything out,” says Mangharam. “There’s always human context—hard-earned domain knowledge, engineering realities, safety rules—that doesn’t live neatly in data and can’t simply be learned from scratch. If we want these systems to work in the real world, we have to teach them the fundamentals we already understand. By building those physical limits, safety boundaries and operational principles directly into the system, we develop physics-informed neural networks, or PINNs, which give AI the necessary domain knowledge on how the world works, the expectations and the lines you can’t cross.”
Read more at
Penn Engineering
.
Share this article
Facebook
LinkedIn
Threads
Credits
Writer
Melissa Pappas
More from
School of Engineering & Applied Science
Artificial Intelligence
Novel plant-based approach to a better, cheaper GLP-1 delivery system
Health & Medicine
Novel plant-based approach to a better, cheaper GLP-1 delivery system
Research led by Penn Dental’s Henry Daniell investigates the use of a lettuce-based, plant-encapsulated delivery platform as a new oral delivery of two GLP-1 drugs previously approved by the FDA in injectable form.
No brain, no gain: Neuronal activity enhances benefits of exercise
Image: Sciepro/Science Photo Library via Getty Images
Natural Sciences
No brain, no gain: Neuronal activity enhances benefits of exercise
Research led by Penn neuroscientist J. Nicholas Betley and collaborators finds that hypothalamic neurons are essential for translating physical exertion into endurance, potentially opening the door to exercise-mimicking therapies.
Studying Shakespeare through the lens of love
In honor of Valentine's Day, and as a way of fostering community in her Shakespeare in Love course, Becky Friedman took her students to the University Club for lunch one class period. They talked about the movie "Shakespeare in Love," as part of a broader conversation on how Shakespeare's works are adapted.
nocred
Arts & Humanities
Studying Shakespeare through the lens of love
In Becky Friedman’s English course Shakespeare in Love, undergraduate students analyze language, genre, and adaptation in the Bard’s plays through the lens of love.
Beating the heat: Designing cooling for bodies in motion
nocred
Technology
Beating the heat: Designing cooling for bodies in motion
Dorit Aviv, director of Weitzman’s Thermal Architecture Lab, studies how humans, technology, and design intersect, paving the way for the development of novel approaches to cooling people efficiently.
Skip to Content
Skip to Content
News from
University of Pennsylvania
Try Advanced Search
What happens when artificial intelligence agents leave the digital world and enter the physical world? Researchers at Penn’s
School of Engineering and Applied Science
are tackling that question through a new three-year, international collaboration on Swarm AI, led by
Rahul Mangharam
, professor in electrical and systems engineering and principal investigator of the
Safe Autonomous Systems Lab
(xLAB). The project brings together three universities to study how large teams of physical AI agents can cooperate, compete, and act safely.
View large image
Ph.D. student Hongrui Zheng (left), and Rahul Mangaharam, ESE professor and principal investigator of xLAB, with the drones used in this research.
(Image: Courtesy of Penn Engineering)
“Most of today’s AI agents live purely in software,” says Mangharam. “We’re moving toward physical AI, systems that don’t just generate answers, but act in the real world. And once AI operates in physical space, it has to deal with real constraints and real consequences.”
Unlike digital agents, physical agents must obey the laws of physics. They must avoid collisions, respect safety boundaries, and coordinate with teammates. The
research
centers on cooperation and coordination in adversarial games—scenarios in which teams of agents must strategize against opponents while maintaining internal cohesion.
“A key technical focus is understanding intent,” says Mangharam. “Agents must infer what other agents, human or machine, are trying to achieve and adjust accordingly. They must coordinate without centralized control and respond to dynamic, uncertain environments. This project brings together research in machine learning for multi-agent systems that use game theory to cooperate and compete.”
At large scales, this becomes an algorithmic and a systems problem: how to design distributed algorithms that scale to tens, hundreds, or even thousands of agents to make consistent, safe decisions in real time. One distinctive aspect of the project is its emphasis on neurosymbolic AI, which combines neural networks with structured, human-encoded knowledge.
“You can’t just throw AI at a problem and expect it to magically figure everything out,” says Mangharam. “There’s always human context—hard-earned domain knowledge, engineering realities, safety rules—that doesn’t live neatly in data and can’t simply be learned from scratch. If we want these systems to work in the real world, we have to teach them the fundamentals we already understand. By building those physical limits, safety boundaries and operational principles directly into the system, we develop physics-informed neural networks, or PINNs, which give AI the necessary domain knowledge on how the world works, the expectations and the lines you can’t cross.”
Read more at
Penn Engineering
.
Share this article
Threads
Credits
Writer
Melissa Pappas
More from
School of Engineering & Applied Science
Artificial Intelligence
Novel plant-based approach to a better, cheaper GLP-1 delivery system
Health & Medicine
Novel plant-based approach to a better, cheaper GLP-1 delivery system
Research led by Penn Dental’s Henry Daniell investigates the use of a lettuce-based, plant-encapsulated delivery platform as a new oral delivery of two GLP-1 drugs previously approved by the FDA in injectable form.
No brain, no gain: Neuronal activity enhances benefits of exercise
Image: Sciepro/Science Photo Library via Getty Images
Natural Sciences
No brain, no gain: Neuronal activity enhances benefits of exercise
Research led by Penn neuroscientist J. Nicholas Betley and collaborators finds that hypothalamic neurons are essential for translating physical exertion into endurance, potentially opening the door to exercise-mimicking therapies.
Studying Shakespeare through the lens of love
In honor of Valentine's Day, and as a way of fostering community in her Shakespeare in Love course, Becky Friedman took her students to the University Club for lunch one class period. They talked about the movie "Shakespeare in Love," as part of a broader conversation on how Shakespeare's works are adapted.
nocred
Arts & Humanities
Studying Shakespeare through the lens of love
In Becky Friedman’s English course Shakespeare in Love, undergraduate students analyze language, genre, and adaptation in the Bard’s plays through the lens of love.
Beating the heat: Designing cooling for bodies in motion
nocred
Technology
Beating the heat: Designing cooling for bodies in motion
Dorit Aviv, director of Weitzman’s Thermal Architecture Lab, studies how humans, technology, and design intersect, paving the way for the development of novel approaches to cooling people efficiently.