Teaching Complex Systems to Navigate

When we’re young, we learn the behaviors that produce positive and negative outcomes. We navigate new environments through trial and error, testing possibilities by gauging those actions that bring rewards and those that bring disaster.

That’s the approach Xiangnan Zhong, Ph.D., brings — only she’ll be using it on autonomous vehicles, not children.

Zhong, who specializes in intelligent control on cyber physical systems — devices like smart energy grids, sensors and interactive games — calls the approach self-learning. It’s a form of reinforcement learning in which a control signal is generated by feedback from the environment.

“Based on the feedback, the system says, this is good. I approach my goal. Or this is bad. I back away from my goal,” Zhong said. “The research is like our human growing-up process, but we transform this process to a machine. We want the machine to grow up like humans grow up themselves.”

Zhong became interested in this field when she was a doctoral student at the University of Rhode Island studying electrical engineering and encountered adaptive dynamic programming — a neurologically inspired approach that allows researchers to optimize control of complex systems. When she accepted an assistant professorship at the University of North Texas in 2017, artificial intelligence was really hot, so Zhong sought to integrate it with her ongoing research.

Zhong is now setting up a lab in the Department of Computer and Electrical Engineering and Computer Science with two awards from the National Science Foundation, including a Computer and Information Science and Engineering Research Initiation Initiative grant, an award that recognizes up-and-coming faculty.

Recently, she received the 2019 Doctoral Dissertation Award from International Neural Network Society, which is presented to the best doctoral dissertation in neural networks, machine learning and related fields.

While Zhong is working on autonomous land vehicles now, she came to FAU because there are many faculty working on autonomous underwater vehicles — her next venue for teaching complex systems how to navigate in our watery world.