How do you get humans and autonomous systems to work together as a cohesive team?  How do you quantify trust between humans and autonomous systems in such teams? For the U.S. Department of Defense (DoD), these are critical questions to be answered.

The Summit on Trusted Autonomy Research & Technology (START) event, held June 28 and 29 at Purdue, was meant to be a “start” towards building a community that would make such trusted human-autonomy teams a reality.   

The summit was hosted by the Purdue Center for Innovation in Control, Optimization and Networks (ICON), in partnership with the Office of the Undersecretary of Defense for Research and Engineering (USDR&E) at the Pentagon; the National Security Innovation Network; and The Johns Hopkins Institute for Assured Autonomy, with key support from Lewis-Burke Associates.

The summit was organized around a set of generative brainstorming sessions where all of the participants contributed ideas towards enabling trusted autonomous systems.

Over 120 experts from Purdue and other top institutions around the U.S., the DoD, research labs, government agencies, and commercialization sectors defined appropriate notions and baseline attributes for operational trusted autonomy, identified end-state goals, characterized current state-of-the-art, and shaped requirements for future investments. START also identified effective strategies for transitioning ideas from the lab into practice through commercialization and workforce development.

The summit opened with remarks by Dr. David Honey, Deputy Undersecretary of Defense for Research and Engineering, who emphasized the critical role that trusted autonomy will play in the DoD’s future missions. “It’s clear that the future of warfare includes autonomous systems,” he said.

The DoD will apply the lessons learned from the summit to successfully support a science and technology strategy for trusted autonomy that will make the country safer and quickly put new, valuable technology into the hands of warfighters.

This multi-year research and development and investment roadmap will enable trusted autonomy for Operational Trust in Mission Autonomy (OPTIMA)  and broader societal applications.

OPTIMA, a DoD initiative identifies the need to deliver trusted autonomy as a product to serve as a battlefield asset in complex, contested missions on the multi-domain battlefield and enable the deployment of effective human-machine teams.

“This is a complex problem," said Shreyas Sundaram, the Marie Gordon Associate Professor in the Elmore Family School of Electrical and Computer Engineering and co-director of ICON. "We need all the stakeholders to weigh in on this problem and start thinking carefully about how to make this real."

Regarding the idea of what it means to "trust" a machine to carry out its intended purposes, Sundaram elaborated on that importance.

"When we ask, 'Do you trust your machine?’ what does that even mean? In order for people, companies, and the government to have trust in these devices, it's not enough to say, 'I trust it.' We need to have numbers associated with it. We need to be able to say, 'Here's why I trust it. Here's how I quantify that trust.' "

This sense of trust between humans and machines is especially important to the DoD due to the critical and sometimes dangerous roles autonomous technology and AI play in war and militaristic settings. The life of soldiers or the sensitivity of vital information can be figuratively or literally in autonomous technology's hands.

A group photo of a subset of attendees at the START summit in June.

In an interview with the Lafayette Journal and Courier, Dr. Jaret Riddick, principal director for autonomy with USDR&E, said that the main future goal with OPTIMA is to deliver trusted autonomy on the battlefield for the warfighter in complex and contested environments. “Industry, academia, and international partners are the folks that we want to engage,” Riddick said. “While the name of the meeting is the Summit on Trusted Autonomy Research Technology (START), this doesn’t imply that we’re starting in an area where there’s nothing.”

While the START initiative was meant to raise awareness of the OPTIMA initiative, work on building and quantifying trust between human and autonomous systems is already well underway. Purdue, and especially ICON, has a particular interest in developing autonomous technology helping to join the efforts of the university with the DoD's OPTIMA efforts.

“Launched in September 2020, ICON aims to address fundamental challenges in autonomous systems by integrating expertise from classical theories in control, optimization and networks with recent advances in machine learning, AI and data science”, said Shaoshuai Mou, associate professor of aeronautics and astronautics and co-director of ICON. Mou is also interim co-chair of Purdue’s Autonomous and Connected System Initiative (ACSI), “We are especially interested in bringing together efforts across academia, government agencies and industry to work on this challenging area of human-autonomy teaming.” 

“ICON, and Purdue in general, is very invested in helping to grow autonomous systems,” Sundaram said. “Autonomous systems are already prevalent, but these complex systems will even be more so. There's a huge amount of research going on at Purdue and within ICON to make these systems be more safe and more reliable. As such, OPTIMA is meant to help transition to a future where we do have trusted autonomy working side-by-side with humans.  We look forward to continuing to be a key partner for the DoD and industry as we work towards realizing this vision.”

At the event, remarks were also given by Mung Chiang, Executive Vice President for strategic initiatives and the John A. Edwardson Dean of the College of Engineering; Theresa Mayer, executive vice president of research and partnerships; and Wayne Chen, associate dean of engineering for research and innovation and the Reilly Professor of Aeronautics and Astronautics & Materials Engineering, highlighting Purdue’s unique capabilities and the key strategic roles that Purdue will play in enabling DoD priorities.