The most vibrant categories of automation are robotic learning. Programming a robot has traditionally required a lot of technical know-how, but what if there was a simpler way for non- programmers to teach the system to do what we want?
Reinforcement learning and imitation are two of the most popular ways to learn. The first requires taking control of the robot to teach it to perform a task and the second requires training a system on millions of images.
A number of researchers are looking at a way to train a system by watching a human complete a task. The in-the-Wild Human Imitating Robot Learning, or WHIRL, is a program that can be trained by watching a video.
TC Sessions: Robotics 2022
In their demos, an off-the-shelf mobile robotic arm learns to open and close drawers and take out the trash.
Shikhar Bahl said thatitation is a good way to learn. It's an unsolved problem in the field, but this work takes a significant step in enabling that ability.
In a home setting, where roboticists anticipate these systems will one day be deployed to help elderly homeowners and other people with impaired movements, it is easy to see how a feature like this would be useful.
Special add-ons are not needed in the case of WHIRL. Even if it takes several times to fully master a task, the robot still tries to execute it until it succeeds. The system is looking to find the best way to complete the task based on its own hardware limitations.
The system is trained by watching videos and the team is looking to add clips from services like YouTube.
All Rights Reserved © 2024
