UA_Engineer_25_FINAL_spreads (1) - Flipbook - Page 11
| FEATURE |
W
hen a child asks for a puppy, they’re
usually not thinking about a robotic one.
But that’s exactly what a student in a
rural Alabama classroom imagined,
and Dr. Chris Crawford was listening.
“We go out to schools a lot,” said Crawford, an
assistant professor in the department of computer
science. “Every time we 昀椀nish our activities,
we have focus group discussions. During those
discussions, we’re primarily trying to 昀椀nd out what
things the students really like and what things they
found challenging.”
The focus groups also include a brainstorming
moment. Crawford invites students to imagine
if they could control anything with their brain or
muscles, what would they want?
“One of the young ladies said, ‘Oh, it would
be cool if we had a puppy robot,’” he said. “Now,
why did she say a puppy robot? I don’t really know,
because she didn’t say a dog robot. That got my
wheels turning because it was the 昀椀rst time I’ve
ever heard a student say puppy robot.”
Though dog-like robots already exist, the student
had been speci昀椀c. She wanted a puppy.
“I got to thinking,” he said, “most of our work
is positioned in rural Alabama. Being born and
raised there, I’m familiar with waking up and seeing
chickens, cattle and pigs in the rural areas. It gave
me the inspiration to really try to integrate the puppy
platform into our work. The project was directly
inspired from one of our students.”
The idea led Crawford to explore the PuppyPi
platform, a quadruped robot developed by
HiWonder that resembles a small dog and can
perform physical actions.
Though the hardware itself came from a thirdparty developer, it was a group of undergraduate
computer science students who brought the project
to life. Under Crawford’s direction, Eli Weber,
Alicia Reed, Olivia Monteiro, Weston Archer Taylor
and Daniel Steuer developed a natural language
processing system that allows users to control the
robot using everyday speech.
“We implemented novel software systems
inside of PuppyPi that allowed it to recognize and
respond to natural language,” Crawford said.
“There’s a system that the students built to connect
that natural language to some of the functions that
PuppyPi has.”
That software allows the bot, affectionately
named “Scuffy” by Crawford’s students, to walk,
turn, sit and ‘box’ on command. It can even be
mischievous, adding personality that resonates with
students and makes the experience more engaging.
Crawford’s research focuses on human-robot
interaction and brain-computer interfaces. Using
Scuffy to teach students topics like robotics and AI
makes STEM education accessible to students in
rural and underserved communities by letting them
speak to the robot in the same way they would talk
to a person.
“This idea of natural language is that I should
be able to talk to PuppyPi in a similar way to how
I’m talking [to a human], and it will be able to
understand it,” Crawford said. “It will be able to
reason around that and then respond in a manner
that’s somewhat accurate or that maps to that
initial request.”
This eliminates the need for users to memorize
code or rigid commands, something that can be a
barrier for students just starting to explore computer
science. Instead, they can begin with what they
already know: how to speak.
“To my knowledge, there’s no other existing
work exploring both natural language and
physiological sensing from the brain and the
muscles with a quadruped format,” he said. “I think
what makes our work different in this context is
our focus on using multiple sensor types in the
context of education.”
The PuppyPi project builds on years of research
for Crawford, who hopes to integrate one of his
existing programs, NeuroBlock, with the robot.
“NeuroBlock is a software that allows students
to build muscle and brain-computer interfaces,”
Crawford said.
RIGHT: Eli Weber watches Scuffy, a PuppyPi robot,
execute a programmed command.
FALL 2025 | CAPSTONE ENGINEER | 11
