Enhanced human-AI collaboration possible through new grant

Erin Cassidy Hendrick
November 12, 2020

UNIVERSITY PARK, Pa. — The Defense Advanced Research Projects Agency (DARPA) has awarded a team of researchers $5.6 million to merge the strengths of artificial and human intelligence, with the goal of advancing the utility of relevant cyber-physical systems. 

The co-principal investigator of the project and the Penn State lead, Christopher McComb, assistant professor of engineering design and mechanical engineering, will receive $437,023 of the grant. His collaborators include Susmit Jha, principal investigator and principal computer scientist at SRI International, and researchers from the University of Texas at Austin.

Every day, the United States military is tasked with countless missions and tasks where both human and computer decision-making skills are critical, such as the deployment of personnel and supplies, unmanned vehicles and the construction of emergency medical facilities. Similar challenges can also be found in businesses and corporations. The sheer amount of data and considerations for security and safety can make these tasks daunting for a human operator, so many systems utilize artificial intelligence (AI) to absorb some of the complexity. 

To significantly improve these systems, McComb and his team seek to streamline and enhance the collaboration of humans and AI through a foundational approach. 

“This gives me a chance to help people live more dignified and interesting lives. The ultimate goal isn’t to create some system that replaces people,” McComb said. “Instead, we want to build something that takes over some of the boring, detailed work and frees humans up to do what we do best – solving big problems.”

To achieve this, McComb and the team of researchers will focus on the creation of a co-designer system, building an AI concept that can generate insights and work with the human operator as quickly and efficiently as possible. Merging these efforts in a collaborative way allows the computer to make smart, data-informed decisions, while allowing the human operator to focus on higher level tasks. 

“We want to create an artificial intelligence system that allows the engineer to navigate multiple objectives as easily as possible,” he said. “Especially to cross between the physical and software domains of cyber-physical systems. We’re trying to train an AI that can solve problems while taking all these domains into account.”

The DARPA program funding this project is the Symbiotic Design for Cyber Physical Systems (CPS), which aims to support AI-based approaches to enable superior designs of military-relevant CPS, in order to adopt innovative techniques and implement them faster into critical operations. 

As one of the participating coalitions, McComb and his fellow researchers will develop the AI co-designer system during the first phase of the project and then, its viability will be tested during a program-wide hackathon. During the first intensive exercise, the team will be designing air taxis. For the next, they will be charged with creating a submarine that can cross under the Arctic’s icy environment autonomously. While these applications are specifically oriented toward challenges the military currently faces, the resulting data will be applied to strengthen the overall algorithmic approach.

“We are aiming to design a framework that can handle multiple objectives, using limited data and across different domains,” McComb said. “It’s something that’s been explored but not fully or at a large scale, so that is where the challenge lies.”

The researchers are hopeful that through this project, they can strengthen and expand the use of human-AI collaboration for designing CPS. 

“What we are figuring out is how can we engineer better systems by focusing on the human element,” McComb said. “The goal is to help people work faster, work smarter, save lives, build better systems, everything.”

 

(Media Contacts)

Last Updated November 16, 2020