As robots become more universal, they will be expected to take on more chores, such as cleaning the house or cooking.
So how do ordinary, non-computer-programming people train their robots? According to researchers at Washington State University, the same way they would train a pet.
The researchers recently presented their work, which used ideas from animal training to assist non-expert users in teaching robots how to perform desired tasks, at the international Autonomous Agents and Multi-Agent Systems conference.
“We want everyone to be able to program, but that’s probably not going to happen,” said Matthew Taylor, Allred Distinguished Professor in the WSU School of Electrical Engineering and Computer Science. “So we needed to provide a way for everyone to train robots—without programming.”
Along with Bei Peng, a doctoral student in computer science, and collaborators at Brown University and North Carolina State University, Taylor designed a computer program that lets humans teach a virtual robot that looks like a computerized dog.
The researchers adjusted the speed that the virtual dogs reacted. So, just like when somebody is teaching a new skill to a real animal, slower movements let the user know that the virtual dog was unsure of how to behave. The user could then provide clearer instruction to help the robot learn better.
Virtual environments in which trainers gave directions to robot dog. (Image Credit: Washington State University)
“At the beginning, the virtual dog moves slowly. But as it receives more feedback and becomes more confident in what to do, it speeds up,” said Peng.
The user taught tasks to the robot by either reinforcing good behavior or punishing incorrect behavior. The more feedback the virtual dog received from the human, the better the robot became at predicting the correct course of action.
The researchers’ algorithm allowed the virtual dog to understand the meanings behind a lack of feedback—called implicit feedback. This happens when you are training a dog, but withhold the treat if the dog does something wrong.
“So no feedback means it did something wrong. On the other hand, when professors are grading tests, they may only mark wrong answers, so no feedback means you did something right,” said Taylor.
Now the researchers are working with physical, real world robots, too. They also hope to eventually use the program to help people learn to be more effective animal trainers.
