Paws and effect: UTC mechatronics team training robot dogs to move like canines

Spot and Goddard took a campus stroll on Friday, Feb. 28. Photo by Angela Foster.

Spot, left, and Goddard took a campus stroll on Friday, Feb. 28. Photo by Angela Foster.

On a recent Friday morning, two robot dogs—Spot and Goddard—roamed across the Chamberlain Field grass as University of Tennessee at Chattanooga Associate Professor of Mechatronics Gokhan Erdemir and two student researchers captured movement data.

Their goal? Refining an artificial intelligence-driven control system to make the robots’ motions more closely resemble those of real dogs.

Erdemir, who teaches in UTC’s College of Engineering and Computer Science, is spearheading research that could advance robotics applications in emergency response, search and rescue, and terrains inaccessible to wheeled robots.

“We’re collecting new data from the robot dogs—their upper and lower legs and body movements—and comparing it to movement data from real dogs,” said Erdemir, a member of the UTC faculty since 2022. “With this information, we can develop a new control algorithm that allows the robots to behave and move like real dogs.”

The research, Erdemir said, has far-reaching implications.

“We want to improve the dexterity and walking abilities of robotic dogs using AI-based control models,” he said. “This could make them more efficient for real-world applications.

“These robotic platforms offer more flexibility than wheeled robots. They can climb mountains, navigate tight paths and reach places that wheeled robots cannot. That makes them valuable for emergency conditions and rescue operations.”

The research team includes UTC mechatronics students Trevor Kight, a senior from Spring Hill, Tennessee, and Ann Mcabee, a junior from Chattanooga.

Kight has been Erdemir’s principal researcher on the project for the past two semesters, focusing on developing code to extract movement data from the robot dogs’ sensors.

“Last year, I worked on coding Spot so we could measure how it moves,” Kight said. “Now, we’re comparing that data with real dog movements so we can create control algorithms that make Spot act and move like a real-life dog.”

Spot, the larger of the two robots, is a Boston Dynamics quadruped. The smaller robot, a Unitree Go2, has been affectionately named Goddard—after the robot dog from the Nickelodeon cartoon Jimmy Neutron.

Mcabee, a nontraditional student who previously studied physics before switching to mechatronics, was assisting with data collection for the first time during the Chamberlain Field excursion.

“This is my first time seeing these robots in action,” she said. “It’s really exciting to learn what they’re capable of.”

Her inspiration to pursue mechatronics came from an unexpected source—her 13-year-old daughter.

“My daughter comes up with crazy ideas and I wanted to be able to implement some of them one day,” Mcabee said. “Mechatronics seemed like the best path because it combines so many engineering disciplines.”

Mcabee joked about telling her daughter about the on-campus research.

“When I go home and tell my daughter I was playing with robotic dogs all day, she’ll probably be terrified,” Mcabee said. “She’s actually scared of robots. But I think they’re adorable.”