Enter a project entitled Deep Transfer Learning Using Depth Sensing of Deformable Fabric - or, "How a robot can make the bed". Such an activity feel systematic, but for those of you who actually make the bed, we know that the bedsheets have their own ways of working.
Researchers at UC Berkeley and the Honda Research Institute have worked with two large robots to "persuade" them how to make the bed. Such was the scientific nature of their research that they initially worked out a formula for how to correctly apply bedsheets.
... and, there are some delightful attempts captured on video. Here's one of the robot moving around the bed to understand where the top blanket needs to be fixed, something which is then successfully carried out.
Here's one of the final bed-making. Well done robot!
The researchers conclude their report as follows - and we honestly can't wait for the follow-up:
This paper uses supervised learning to teach mobile robots to make a bed across multiple blanket configurations and colors.
There are a number of future work directions, such as extending the method for application to furniture covers, table cloths, and other deformable objects commonly seen in homes. An additional extension is to develop policies that can pull out and reveal hidden or unreachable corners.