ADVERTISEMENT
ADVERTISEMENT
UPenn's Rahul Mangharam advances Swarm AI for real world interactions
The Indian-origin professor highlighted that unlike digital AI, physical AI systems need to be taught fundamentals of physics.
-
1759953093.png)
Staff Reporter - Wed, 15, Apr 2026
- 04:04 EDT
- 18
Rahul Mangharam / Rahul Mangharam via LinkedIn
Prof. Rahul Mangharam of the University of Pennsylvania is leading research on Swarm AI, a new initiative focused on how large teams of physical AI agents can cooperate, compete, and operate safely in real-world environments.
Mangharam, along with fellow researchers at Penn’s School of Engineering and Applied Science, is studying how distributed algorithms can coordinate tens, hundreds, or even thousands of agents in real time without centralized control. The work is part of a new three-year international collaboration bringing together multiple universities.
The Swarm AI project aims to move beyond traditional software-based systems toward physical AI, where machines actively interact with the real world.
“Most of today’s AI agents live purely in software,” Mangharam said. “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.”
ALSO READ: Anantharam Raghuram named Simons mathematics fellow
Unlike digital AI, physical AI agents must obey the laws of physics, avoid collisions, respect safety boundaries, and coordinate with other agents in dynamic environments.
“A key technical focus is understanding intent,” Mangharam said. “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.”
The research combines machine learning with multi-agent systems and game theory, enabling agents to both cooperate and compete effectively.
Mangharam emphasized that physical AI cannot rely solely on learning from data. Instead, systems must be built with embedded domain knowledge.
“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,” he said.
Discover more at New India Abroad.
Related
ADVERTISEMENT
ADVERTISEMENT
E Paper
Video
Comments
Start the conversation
Become a member of New India Abroad to start commenting.
Sign Up Now
Already have an account? Login