October 18, 2022
By Ian Sargent

George W. Woodruff School of Mechanical Engineering Assistant Professor Ye Zhao was announced as one of 25 recipients of the 2023 Office of Naval Research’s (ONR) Young Investigator Award. The ONR’s Young Investigator Program targets early-career scientists whose research carries significant potential for technological advancement and national defense support. Zhao’s proposal, selected out of a pool of nearly 200 applicants, aims to dramatically improve the way autonomous robots make sense of and navigate complex and fast-changing environments.

“I’m honored and grateful to be selected for this award,” Zhao said. “This is a big recognition not just for me, but for the whole lab and talented students. It helps further promote our research effort in humanoid robot locomotion and manipulation, and enables us to picture a broader research scope.”

Zhao directs the Laboratory for Intelligent Decision and Autonomous Robots (LIDAR) at Georgia Tech, which focuses on the development of collaborative robots and improving their decision-making and planning capabilities.

Industrial robots have been widely used in some form or another since the 1950s, primarily in large-scale fabrication, where their ability to perform heavily repeated tasks provided revolutionary support to manufacturing industries. While these robots have grown nimbler and more sophisticated over time, their superiority to manual labor is still limited to structured environments that are not expected to change—ideal for assembly lines or assisting in laborious industrial processes. When placed in a dynamic real-world setting, an industrial robot would lack the necessary adaptability to perform a diverse range of tasks successfully.

The robots Zhao and his team at LIDAR build differ from industrial robots in versatility and morphology. As humanoid robots, they, like us, can potentially perform complex sequences of tasks in a vast range of settings and conditions.

“In general, humanoid robots can have a broader application scope,” Zhao said. “From agricultural fruit harvesting to construction sites, to home assistance. Humanoid robots have great potential to do whatever humans can do, go wherever humans can go.”

Zhao’s proposal targets naval search and rescue scenarios, where a robot would need to work in confined maritime spaces with or near humans. This calls for a design that is both strong enough to withstand hazardous environments and safe enough to be deployed near humans. To do this, Zhao will design a series of advanced algorithms that will allow a humanoid robot to sense and safely navigate its environment.

“One of the main challenges stems from the computational hurdles that prohibit these complex-legged systems from operating in unstructured environments reliably,” Zhao said. “This project will design intelligent algorithms we call ‘contact-aware trajectory optimization and decision-making’ with real-time safety guarantees for versatile locomotion and manipulation across a wide breadth of terrains and environments with humans in the scene.”

The novelty in Zhao’s algorithm is in dividing the problem-solving tasks into two levels working simultaneously. The first, or low level, will continuously take in information from the surrounding environment like terrain, obstacles, and other humans, and use the data to guide the robot’s movements. The second, high-level process handles overall task planning and decision-making.

Zhao’s proposal is part of the ONR’s Human Interaction with Autonomous Systems program, which seeks to develop autonomous systems that can work alongside humans and other robots. This closely mirrors Zhao’s long-term career goals.

“If we could solve some of these challenges, I believe there will be a great chance for humanoid robots to be deployed for real-world tasks,” Zhao said. “Seeing humanoid robots in our daily life would probably have a long way to go, probably five years or more.”