Large unmanned surface vessel Ranger transits the Pacific Ocean. (Credit: US Navy Photo)

The Office of Naval Research (ONR) has selected Purdue University to partner with Saab on the development of next-wave artificial intelligence capabilities to increase battlespace situational awareness by automatically detecting and characterizing threats in complex environments.

The Threat and Situational Understanding with Networked-Online Machine Intelligence (TSUNOMI) program will capitalize on Saab’s growing presence in West Lafayette, leveraging its advanced manufacturing facility and engineering resources. While the technologies developed through TSUNOMI are first intended for transition to a naval unmanned surface vessel, they are extensible to other markets, such as military radar systems and commercial airport applications that are complementary to Saab’s growing portfolio in the United States, according to a Saab release.

The total grant from ONR to Saab is approximately $13 million, of which an anticipated $4.3 million will filter to Purdue as a four-year subcontract through Saab.

Co-principal investigators for the project are Shaoshuai Mou, Elmer Bruhn Associate Professor of Aeronautics and Astronautics, and Shreyas Sundaram, the Marie Gordon Associate Professor in the Elmore Family School of Electrical and Computer Engineering. Both professors serve as co-directors of Purdue’s Institute for Control, Optimization and Networks (ICON), launched in 2020.

“This award is a great example of highly impactful research that can result from ambitious and collaborative faculty leadership and is supported by Purdue Engineering initiatives in autonomous connected systems, national security and technology, and hard AI,” said Arvind Raman, the John A. Edwardson Dean of the College of Engineering.

TSUNOMI comes at a time when serious physical attacks on critical infrastructure and intrusions into sensitive areas, both in military and civilian domains, are increasing, Sundaram said.

“There is an urgent need to create technological solutions that allow networks of sensors equipped with sophisticated AI to quickly detect and identify potential threats,” he said.

ICON, which includes 84 faculty from across the College of Engineering, College of Science, College of Agriculture and Purdue Polytechnic Institute, was a natural choice for Saab, Sundaram said.

“ICON’s rapid growth, expertise and significant level of activity make it a top-tier center for autonomous systems research and education. The breadth and depth of expertise within ICON also allows us to quickly pull together key experts to form effective teams to tackle new projects,” Sundaram said.

The objective of the TSUNOMI research is to develop an explainable machine learning framework with multimodal automatic target recognition and sensor resource management for early warning and situational awareness from surface vessels equipped with an automated verification and validation machine learning pipeline, Mou explained.

“This partnership is a major recognition of ICON’s leadership efforts in frontier research in autonomy,” Mou said. “On one hand, I was very glad to hear that we received the award, since the project aligns very well with ICON’s scope and is of significant research impact. On the other hand, I was not very surprised, since the ICON-Saab team has a lot of preliminary results in multi-agent autonomy and the application of machine learning algorithms in situational awareness.”

A group of ICON faculty is partnering with Saab on a smaller project funded by DARPA (the Defense Advanced Research Projects Agency).

TSUNOMI is ICON’s first large initiative with Saab and under support from the Office of Naval Research, Sundaram said.

“The project seeks to formulate effective techniques and algorithms to blend information from multiple sensors, such as cameras and radar, that have been deployed in an area to accurately identify objects that might enter that area,” he said. “For example, one might wish to quickly determine if a flying object near an airport is a bird or a drone, and if it is a drone, whether or not it is adversarial and what its target might be.”

"TSUNOMI is a great win for the Saab and Purdue partnership,” said Erik Smith, president and CEO of Saab. “This program represents a real step toward robust and trusted artificial intelligence. TSUNOMI will help stakeholders make decisions quicker and with more confidence."

In this scenario, there are a host of challenges to solving these problems, Sundaram said, including where to deploy the sensor for adequate coverage, how to create machine learning algorithms to identify the object and how to blend information from multiple sensors to accurately determine the type and intent of the object.

To address the challenges, Purdue investigators will bring expertise in a core area of the overall problem:

• Mou will work on algorithms for intent classification and threat evaluation.
• Sundaram will work on a sensor resource management engine that will guide the appropriate selection and deployment of sensors.
• Dan DeLaurentis, aeronautical and astronautical professor and vice president for research institutes and centers, will serve as technical advisor to the team, bringing his expertise in systems-of-systems analysis and design, as well as decades of experience in delivering effective solutions for military applications.
• James Goppert is the managing director of Purdue UAS Research and Test Facility, where key experiments will be performed and the algorithms will be validated.
• Christopher Brinton, Elmore Assistant Professor of Electrical and Computer Engineering, will lead the development of federated learning algorithms, which controls how knowledge gained from individual AI systems is aggregated and shared. He will also serve as a technical lead for the project.
• David Inouye, ECE assistant professor, is an expert on explainable AI and will contribute to ensuring that the outputs of the target recognition algorithms are interpretable by humans.
• Michael Zoltowski, the Thomas J. and Wendy Engibous Professor of Electrical and Computer Engineering, is a leading expert on radar systems and will contribute to ensuring that critical sensors can be integrated into the overall TSUNOMI architecture.
• Song Zhang, assistant head for experiential learning and mechanical engineering professor, will provide expertise on sensor fusion and tracking algorithms.
• Carl Huetteman, managing director of special initiatives for the College of Engineering, will serve as project manager for the team, coordinating with Saab and ONR to keep deliverables and outcomes on track.

Purdue’s funding will primarily be used to support graduate students on the project, as well as some faculty time. Part of the award also will be used to purchase sensors.