At the time of her interview, Smith was in the final stages of her project exploring new applications for one of the lab’s more underutilized robots. When the sounds of tumbling blocks and shrieking robots finally subsided, Smith returned to explaining her summer research. “Have you heard of Cozmo?” she asked.

Cozmo is possibly the cutest robot you’ll ever meet. Big-eyed and pint-sized, the wheeled robot raced across a lab table according to Smith’s directions, drawing out the patterns of diamonds, circles, and squares.

“I want to use him as an assistive device for children with cerebral palsy,” she said as the robot accelerated away from her. “I’m exploring ways he can be programmed with accessible languages like Scratch. Cozmo’s cute for a reason. He’s supposed to encourage motivation and engagement. And I think he can help patients interact, especially kids who are non-verbal.”

“Look at us,” Smith laughed. “We’re even talking about Cozmo as if he’s someone we know.”

But that’s exactly the goal of the Rehabilitation Robotics Lab, which studies human-robot interaction to develop the most effective therapeutic robots for neurorehabilitation. By investigating the features of robots that make them more appealing — big eyes, a firm handshake, or some great dance moves — they can build a robot that’s better at its job of healing others.

Smith’s RET colleague, sixth and seventh grade math teacher David Dimatties, was busy working on those dance moves. His project aimed to hack My Keepon, a socially assistive, dancing robot, to use as a motivational tool in physical therapy.

“I’m trying to hack this robot to get increased functionality to motivate and engage cerebral palsy patients,” he said, typing commands in Python on his laptop, “but it’s not working.”

He’d been trying to hack the bouncy yellow bot for over a week with no luck. Frustration can be a familiar feeling for the teachers in Penn’s RET program, where they’re often thrown into the deep end of Penn’s latest research.

“Nobody’s stuff ever works quite the way you expect it to,” noted Michael Sobrepera, Smith and Dimatties’ mentor and an Engineering Ph.D. student in the Rehabilitation Robotics Lab. “I think it’s really hard to understand how hard research is in the general scope until you really try to dig into it.”

“You’re doing things that by definition haven’t been done yet,” he continued. “Which means that every time you hit a roadblock, there’s not a YouTube video to watch or a book to read. You have to figure it out, because you’re going to be the one writing about it. And that’s challenging.”