Bookmark and Share

Students explore human-robot interaction

By Thao Nguyen ’15

In a field so new that there are no formal textbooks available, 20 students in Ross Sowell’s Human-Robot Interaction course relied on the latest research as a guide. Drawing from a range of disciplines that included computer science, psychology, and engineering, the students applied their newfound knowledge to their own projects using Turtlebot mini-robots.

Students pitched ideas and eventually settled on six projects: a robot guide dog, a robot that can solve mazes, a robot bartender, a robot security guard, two robots that can play tag with each other, and a system to protect data privacy by blurring certain images viewed by a robot.

Students from the guide dog project created a robot that moved according to an order from its owner, and that would then alert the user if there was an obstacle in its way. A robotic dog could replace guide dogs in assisting blind people, and the project involved more than just programming a robot that respond to voice commands.

“We tried to understand the perspective of blind people in order to design a system that allows them to feel comfortable with the robot,” said sophomore Chell Robison. For example, their robot could participate in casual conversations, and they always kept it at a distance from the owner so he/she would not feel threatened.

Sowell spent the last summer researching human-robot interactions at Oregon State University, and decided to introduce the topic this year at Cornell.

“Because this is such a young field, it offers a lot of opportunities for undergraduate research,” he said. “Students don’t have to gain a lot of prior knowledge in order to make significant contributions to the field.”

Junior Alex Hubers worked with Sowell last summer, and in this class he furthered his work on the topic of privacy protection on tele-operated robots. As an example of this application, a person might grant control of their robot to a remote operator, such as a plumber. The goal is to protect privacy by limiting what the robot can “see.”

“Besides programming, we needed to design a plan and manage our project, including what we wanted to achieve by what deadlines,” Huber said. “This class gave me a clear sense of how research and real life work are done.”