Zachary Sunberg News

Professor earns two major grants to advance AI for autonomous systems

Jeff Zehnder — Tue, 03 Sep 2024 14:24:05 +0000

Professor earns two major grants to advance AI for autonomous systems Jeff Zehnder Tue, 09/03/2024 - 08:24

Jeff Zehnder

Zach Sunberg

Zach Sunberg’s research developing better artificial intelligence systems is getting a major boost from two federal grant awards.

Sunberg is receiving a $599,000, five-year CAREER Award from the National Science Foundation and is a partner on a related $4 million multi-university initiative from the U.S. Office of Naval Research.

Both projects focus on advancing game theory algorithms so AI systems can better solve problems in real-time in the field.

“I’m excited to receive the recognition that I’m looking at important problems to solve in both of these areas,” Sunberg said. “ONR shows the relevance for defense applications, and the NSF award focuses on making our nation and our world a better place.”

Sunberg is an assistant professor in the Ann and H.J. Smead Department of Aerospace Engineering Sciences at the University of Colorado Boulder. His research focuses on autonomous systems and AI, with an emphasis on game theory.

Although game theory has origins in solving tabletop board games and card games, it is a broad field of research that studies problem-solving in complex, real-world situations.

“You might think of a game as a board game, but any situation where multiple agents are interacting and have their own goals can be mathematically represented as a game. Poker has clear rules, but so does driving a car; there are just a lot more variables,” Sunberg said.

AI systems like those used in self-driving cars typically rely on offline reinforcement learning. In such a system, automakers use historical data collected from a fleet of vehicles to optimize an algorithm to react to future situations. Sunberg is seeking to develop online decision-making systems, where an AI can think in real time to interact with situations that do not match historical data.

“This has previously been considered a computationally intractable problem” Sunberg said. “But our lab recently had a breakthrough with single-agent planning where we proved we did not need a lot of computation even in a large state space. What we want to do next is work on more complex multi-agent situations.”

The research focuses on a framework used by scientists and engineers to model possible outcomes when full data is not available, called partially observable Markov stochastic games.

“An application is airborne collision avoidance. In the past, the other pilot’s actions would be modeled as a probability distribution. That wasn’t satisfying to me. The other pilot is a decision maker themselves, so it would be better to model as a multi-agent game, but we don’t know how to solve partially observable games like that using online systems,” Sunberg said.

The research from the NSF grant has applications across an array of autonomous systems, from search and rescue robots, to self-driving cars, to how satellites navigate while orbiting the Earth. The Navy award is focused more on AI deception and counter deception in the military realm.

“An enemy is going to try to deceive you in some way, so we want to focus on how that can happen and how do we make AI resistant to it. We’re also looking at developing AI that can deceive an adversary. If you have a drone that you want to avoid enemy air defenses, how can it use bluffing to help it do that,” Sunberg said.

The ONR award is brings together four universities. The project is being led by the Georgia Institute of Technology with partners at the University of California Santa Barbara, ��ý�Ļ��Ʒ, and the University of Texas at Austin. Sunberg’s portion of the $4 million grant is worth roughly $1 million.

“The current most widely used methods for AIs, these offline systems, are really a function approximater. It’s kind of an intuitive reaction or instinct. We want an AI that can go further, like people, and think and deliberate about a situation. There’s huge potential with this work,” he said.

Zach Sunberg’s research developing better artificial intelligence systems is getting a major boost from two federal grant awards. Sunberg is receiving a...

Off

Traditional

White

Seminar - Breaking the Curse of Dimensionality in Decision-Making for Autonomous Systems - Sept. 6

Jeff Zehnder — Tue, 03 Sep 2024 13:35:18 +0000

Seminar - Breaking the Curse of Dimensionality in Decision-Making for Autonomous Systems - Sept. 6 Jeff Zehnder Tue, 09/03/2024 - 07:35

Zach Sunberg

Zach Sunberg
Assistant Professor, Smead Aerospace
Friday, Sept. 6 | 10:40 a.m. | AERO 114

Abstract: Autonomous cyberphysical systems hold the promise to positively transform many tasks, for example transportation, collecting scientific data in hazardous environments, and monitoring objects in space. Uncertainty is a critical factor in all of these domains. The partially observable Markov decision process (POMDP) and partially observable stochastic Game (POSG) provide mathematical formalisms for computing the best single-agent and multi-agent policies in the presence of uncertainty, however, these problems are notoriously computationally difficult to solve.

Recent work in the Autonomous Decision and Control Lab (ADCL) at ��ý�Ļ��Ʒ has shown that one source of intractability, namely the curse of dimensionality in the state and observation spaces, is possible to overcome, both theoretically and practically. This presentation will give an overview of the ADCL's efforts to develop new theory and algorithms to solve POMDPs and POSGs and deploy the algorithms to solve real-world challenges.

Bio: Zachary Sunberg is an Assistant Professor in the Ann and H.J. Smead Department of Aerospace Engineering Sciences and the University of Colorado Boulder.

His research focuses on partially observable Markov decision processes and game theory with applications to artificial intelligence and aerospace vehicle control.

Sunberg earned his PhD in Aeronautics and Astronautics at Stanford University, and his BS and MS degrees in Aerospace Engineering from Texas A&M University.

Off

Traditional

White

Panel and Q&A - Charting Your Course: Navigating Careers in Aerospace - Sept. 6

Anonymous — Tue, 29 Aug 2023 20:33:13 +0000

Panel and Q&A - Charting Your Course: Navigating Careers in Aerospace - Sept. 6 Anonymous (not verified) Tue, 08/29/2023 - 14:33

Wednesday September 6, 2023
5-6:30 p.m. • AERO 120

Smead Aerospace invites you to attend the inaugural event of a new workshop series on the broad range of opportunities for careers in aerospace engineering.

We have assembled a panel of three aerospace professionals who will discuss exciting and innovative accomplishments in our field, share experiences from their careers, and provide advice and guidance to your own path.

This will be a one-hour panel with questions from the moderator and audience, followed by an open house that gives you the opportunity to interact with panelists one-on-one.

Come learn about the amazing and exciting career paths of our three panelists:

Off

Traditional

White

��ý�Ļ��Ʒ wins major Air Force grant to track objects orbiting the moon

Anonymous — Mon, 21 Aug 2023 15:25:23 +0000

��ý�Ļ��Ʒ wins major Air Force grant to track objects orbiting the moon Anonymous (not verified) Mon, 08/21/2023 - 09:25

Jeff Zehnder

We want to advance civil space traffic management and planetary defense for asteroids."

The University of Colorado Boulder is leading a major Air Force project to track objects orbiting near the moon.

The Air Force Research Laboratory is awarding a Space University Research Initiative worth up to $5 million over five years to the multi-university ��ý�Ļ��Ʒ-led team. Also collaborating on the research are Texas A&M University, the Georgia Institute of Technology, and L3Harris Technologies.

The goal is to improve space domain awareness in high Earth orbits – above geosynchronous – and in the vicinity of the moon, according to Marcus Holzinger, an associate professor in the Ann and H.J. Smead Department of Aerospace Engineering Sciences and the principal investigator for the grant.

“We want to advance civil space traffic management and planetary defense for asteroids, and assist the U.S. Space Force with space tracking,” Holzinger said. “This is truly a dual use that improves timeliness and the quality of decisions as a benefit to humanity writ large.”

An additional ��ý�Ļ��Ʒ partner on the project is the university’s Center for National Security Initiatives. NSI will facilitate sponsor events and engagement, monitor cost-performance objectives, and identify adjacent defense opportunities to further advance the research and expand its national security footprint.

The project presents unique and complex challenges for all members of the team, said Zach Sunberg, an assistant professor in Smead Aerospace who is also a team member.

“Space is not an intuitive domain. We will need to use innovative algorithms, computational techniques, and experimental research to make this happen. This really brings together experts in areas ranging from orbital dynamics, to human computer interaction, to high tech sensors,” Sunberg said.

NASA and the Air Force currently maintain systems to track asteroids and man-made objects in Low Earth Orbit for military readiness and to prevent collisions of satellites but have more limited capabilities for monitoring objects around the moon.

Up until very recently, that has not been a problem as hardly any missions orbited the moon. The recent explosion of commercial and government projects at higher orbits makes the current system unsustainable. Unfortunately, existing tracking systems cannot be easily adapted for the purpose.

“The current systems are all legacy technology. The algorithms that underpin them depend on Keplerian physics, two-body motion – the Earth and one other body. The region we’re talking about is heavily affected by the gravity of the moon and the sun as well, and the dynamics we use to predict movement begin to fail,” Holzinger said.

The grant will include fundamental and applied research joining telescopes, sensor technology, and artificial intelligence to detect, characterize, and track spacefaring objects in high orbits to improve decision making for humans on the ground.

“We need to know is this a spacecraft or a natural object and where is it going,” Holzinger said. “This improves safety for civilian and commercial space flights and can help us spot asteroids that may be on a collision course with Earth.”

Sunberg said the project has real potential for positive, long-lasting impacts.

“Humanity has the potential to harness the space around the Earth and the moon for a variety of benefits. However, as this area becomes increasingly occupied with satellites controlled by users with different goals, it is more and more important to maintain our situational awareness so it is a safe and useful environment for all. This work will develop technologies critical to that goal,” Sunberg said.

Additional researchers on the grant include Scott Palo at ��ý�Ļ��Ʒ; Terry Alfriend, Kyle DeMars, and John Junkins of Texas A&M University; and Karen Feigh of Georgia Tech.

The University of Colorado Boulder is leading a major Air Force project to track objects orbiting near the moon. The Air Force Research Laboratory is awarding a Space University Research Initiative worth up to $5 million over five years to the...

Off

Traditional

White

Designing flying AI systems to study supercell thunderstorms up close

Anonymous — Wed, 08 Dec 2021 21:40:14 +0000

Designing flying AI systems to study supercell thunderstorms up close Anonymous (not verified) Wed, 12/08/2021 - 14:40

Jeff Zehnder

Above: Members of the ��ý�Ļ��Ʒ flight crew working on a RAAVEN drone in 2019 during a mission with a tornado in the distance.
Header: The RAAVEN drone flying near a storm system.

A team of University of Colorado Boulder scientists and engineers have landed a major grant to design next-generation uncrewed aircraft systems (UAS) to fly into the heart of supercell thunderstorms that can spawn tornadoes.

“We’re trying to improve forecasts of severe weather,” said Eric Frew, a professor of aerospace engineering sciences. “There are things you can’t understand without flying into the storm from the air, things ground radar does not tell you.”

Frew’s team has earned a three-year, $1.5 million National Science Foundation grant to get closer than ever before to the heart of supercell storms: by launching probes that can fly directly inside them.

“We’re flying drones at the edge of storms and are being successful, but we can’t pierce the heart of the storm with the drone because the winds and precipitation are too strong,” said Frew, who is the principal investigator of the grant.

The goal of the research is to develop a drone that can fly as close to a storm as possible and then deploy a series of helium balloon probes carrying sensor packages that will be sucked into the center of the storms at altitude and report back data on the conditions inside.

While UASs are often remote controlled by humans on the ground, the goal of this system is for it to work fully autonomously. The UAV will make its own decisions about how close it can safely get to a storm and when it should release the probes, using artificial intelligence to make choices based on quickly changing storm conditions faster than a team of people could.

Developing this type of system is a new frontier in artificial intelligence research. Assistant Professor Zachary Sunberg is a co-principal investigator on the grant; his expertise is in high-performance artificial intelligence algorithms for decision making under uncertain conditions.

“The storms that we plan to study are extremely difficult to model, so there is a lot of uncertainty,” Sunberg said. “My role will be to use artificial intelligence to optimize the aircraft's path to deploy the balloons in the places that they will gather the most information about the storm. It is one of the most exciting applications of my artificial intelligence work.”

These systems will eventually be deployed in the field during storm-chasing campaigns. Frew and colleagues at ��ý�Ļ��Ʒ have made numerous multi-week excursions across the Great Plains following supercell storms and flying UAVs to gather data that can be analyzed by scientists to improve weather predictions and early warning systems for tornadoes.

“This has all been part of a 15-year vision of an autonomous airborne scientist. It used to be that being in the field was a barrier to communication. Now we have aircraft that have LTE and access to the internet, which opens us up to cloud computing. It makes for much more capable artificial intelligence,” Frew said. “The aircraft is just one part though. The science gathering is the aircraft, the team, the ground station computer, algorithms in cloud servers. We have the big brain of the internet at our disposal in ways we never did before.”

In addition to Frew and Sunberg, the research also includes Brian Argrow, professor of aerospace engineering sciences at ��ý�Ļ��Ʒ, and Adam Houston of the Department of Earth and Atmospheric Sciences at the University of Nebraska-Lincoln.

Off

Traditional

White

Welcoming four new faculty to Smead Aerospace

Anonymous — Mon, 26 Aug 2019 16:41:06 +0000

Welcoming four new faculty to Smead Aerospace Anonymous (not verified) Mon, 08/26/2019 - 10:41

The Ann and H.J. Smead Department of Aerospace Engineering Sciences is welcoming four new faculty members. Meet the team and see why we're so excited about these talented new hires:

Off

Traditional

White

Zachary Sunberg News

Professor earns two major grants to advance AI for autonomous systems

Off

Seminar - Breaking the Curse of Dimensionality in Decision-Making for Autonomous Systems - Sept. 6

Off

Panel and Q&A - Charting Your Course: Navigating Careers in Aerospace - Sept. 6

Off

����������ý�Ļ���Ʒ wins major Air Force grant to track objects orbiting the moon

Off

Designing flying AI systems to study supercell thunderstorms up close

Off

Welcoming four new faculty to Smead Aerospace

Off

��ý�Ļ��Ʒ wins major Air Force grant to track objects orbiting the moon