Abdelzaher’s IoBT paper, outlining future automation of defense with AI, among most downloaded at Army conference

6/18/2020 Allie Arp, CSL

Written by Allie Arp, CSL

The University of Illinois at Urbana-Champaign’s Internet of Battlefield Things (IoBT) alliance – a $25M Army-funded effort that is investigating how machine intelligence can advance future defense – is being led by CSL faculty member Tarek Abdelzaher. A presentation he gave this spring summarizing all that is being done in the UIUC-led alliance has become the second most downloaded technical presentation from the all-virtual 2020 Defense and Commercial Sensing Conference on Artificial Intelligence and Machine Learning for Multi-Domain Operations Applications II.

“A multi-domain operation is when forces from all military domains – land, air, sea, cyber, and satellite – work hand-in-hand to accomplish a joint goal against a near peer adversary,”

Tarek Abdelzaher
Tarek Abdelzaher

said Abdelzaher, professor and Willett Faculty Scholar in computer science. “AI is believed to be the enabler – it’s the technology that will decide who outmaneuvers their competition in a future conflict.”

Abdelzaher’s paper, “The Multi-Domain Operations Effect Loop: From Future Concepts to Research Challenges,” was co-authored by Adam Taliaferro of the Army’s Future and Concepts Center and Paul Sullivan and Stephen Russell of the CCDC-Army Research Laboratory (ARL). The paper offers a way of thinking about AI services for the battlefield and explains the exciting things they make possible.

“Coordinating the thousands of parts in a battlefield is a complex problem. It’s a big loop of observations and reactions,” said Abdelzaher. “We break it down into stages and employ different types of AI at each stage, supported by results from our project. It allows both researchers and military stakeholders to understand which piece of research coming from the labs is best suited for which operation needed in the field. This marriage of research accomplishments and practitioner needs is the main value of our paper.”

One particularly intriguing proposition in the paper is the use of AI and machine learning to improve the military operations decision loop. For civilians, decision-making often involves figuring out a situation, analyzing information, making a decision, and then acting on the decision. Military operations are very similar, but also involve localizing and tracking, along with a higher level of information gathering. This process in the military can take hours or even days to do manually. Using computers and the AI-driven decision “loop” developed by Abdelzaher and his team, the process could be automated and take only minutes.

“Automated defense is important because in times of conflict you need to act fast,” said Abdelzaher. “In some situations, whoever acts the fastest on the best data wins and that happens at a time scale that humans can’t keep up with.”

In taking civilian innovations to the battlefield, this technology will bring the critically important ability to integrate together systems from multiple domains. Currently, the interactions between military organizations are manual and time-consuming.

“In the past, the various domains, land, sea, air, space, and cyberspace, would operate separately,” said Abdelzaher. “We’re working on ways to integrate all the armed forces to operate together, which will take a high degree of organization and networking.”

The research may have applications across the whole spectrum of military activities.

“In addition to improving military activities, AI and IOBT research is fundamentally about improving decision-making,” said Russell, Chief of the Information Sciences Division of ARL. “The foundational innovations from the IoBT CRA, both in terms of AI and transformative command and control, can only increase their impact by being investigated under multi-domain military decisions-making contexts, such as those defined in the MDO Effect Loop concept.”


Share this story

This story was published June 18, 2020.