By Siddharth Vodnala, Voiland College of Engineering and Architecture
PULLMAN, Wash. – Daylan Kelting started programming computers in middle school. He joined a robotics club in high school. In college he got interested in teaching machines to learn.
This summer Kelting is doing research that brings together those interests and skills at the Center for Advanced Studies in Adaptive Systems (CASAS).
Kelting is working under the direction of Diane Cook, professor, and Aaron Crandall, clinical associate professor, in the School of Electrical Engineering and Computer Science.
A rising senior majoring in computer science, Kelting is part of a three-member student team programming an elder-care robot called Robot Activity Support (RAS).
Elder-care robots are being designed and produced all over the world to provide companionship to older adults, coach them during exercise and provide physical assistance like fetching food and turning lights on and off.
RAS is a combination of a coach and physical assistant robot being designed at CASAS to provide in-home support to elders. It tracks the daily activities of older residents and assists them when they need help, like reminding them to take their pills or showing them where the keys are.
Kelting’s team’s task for the summer is to write a computer program to instruct the robot to find the best locations around the house to wait at so that it can reach the residents quickly in case they need help.
The robot uses sensors located inside a “smart home” environment developed by Cook’s lab to build a map of its surroundings.
Kelting’s team is trying to instruct the robot to use the sensor data and the map to navigate around the house so that it is able to wait close to where it is needed.
“I’ve always wanted to do research with robotics and deep learning,” said Kelting, explaining his decision to do summer research at WSU. Deep learning refers to a subfield of artificial intelligence in which researchers use data to train computers to do complex tasks.
Kelting explained that the difference between being in class and doing research is that in class students know there is a right answer to a question.
“In research the questions are open-ended,” Kelting said. “You can put in all the effort in the world and your solution can fail. But the stuff you tried probably serves as a good platform for trying the next thing.”
Kelting, who started his summer research in early June, is not only flexing his programming muscles writing computer algorithms but also acting as project manager for his team.
“I love that I get to learn a lot about managing a project,” he said.
Kelting also got to learn something else soon after starting his research at the lab: the value of a good advisor.
“Discussing our ideas with doctor Crandall gave us a much more solid idea of what we wanted to do,” he said.
Kelting, who is from Lynnwood, Wash., was president of the robotics club at his high school.
After starting college at WSU he approached Matthew Taylor, assistant professor of computer science with his interest in artificial intelligence and deep learning. Taylor pointed him to some online courses and helped with research ideas.
That collaboration led Kelting to research a field of machine learning called reinforcement learning last summer with Taylor at the Intelligent Robot Learning lab.
Kelting is also currently president of the Palouse RoboSub club, an organization focused on building an autonomous underwater vehicle to participate in a yearly international competition in San Diego, CA.
Kelting’s advice to students interested in research is to approach a professor with their interests just like he approached Taylor.
“Finding a good project to work on is important,” he said. “Keep learning more about your field and visit the labs you’re interested in.”
By the end of this summer, Kelting hopes to produce a research paper detailing their algorithm.
“I haven’t written an actual research paper before,” said Kelting, adding that he was “kind of intimidated” by the prospect. But he is confident that his team will be up to the task.
Kelting eventually plans to do a Ph.D. in reinforcement learning, something he said he wouldn’t have considered before he discovered his passion for research.
“The great thing about research is there is no limit to what you can achieve,” he said. “The ceiling is infinite.”
- Brett Stav, public relations/marketing director, Voiland College of Engineering and Architecture, 509-335-8189, email@example.com