Pieter Abbeel is lazy when it comes folding laundry. He’d rather a robot do the dirty work.
That’s how it seemed anyway when I visited his University of California, Berkeley lab. Abbeel stood by his obedient robot and we watched as it slowly folded towels and socks in an orderly fashion.
The robot wasn’t built entirely from scratch, it actually was given to Abbeel by a company called Willow Garage. The bot is part of the PR2 program, an open source software platform that encourages sharing within the robotics community.
In the future, robots might be programmed to perform other chores that we hate to do around the house: cleaning, organizing, unloading and loading dishwashers, cooking and setting up the dining room table.
I spoke to Abbeel to find out more about his domestic robot and to hear what is in store for the future of robotics:
SmartPlanet: So watching the robot fold laundry was as fun as watching paint dry. Why did you build a robot to fold laundry? It seems a little trivial.
PA: We decided to pick laundry for three reasons.
First, robotics have been successful in structured environments such as manufacturing halls, where repeated execution of the same motions is sufficient for automation. In unstructured environments, robots have been far less successful. Doing laundry required interaction with deformable objects.
Second, we work with surgical robotics. It turns out, a robot designed to perform surgery has to handle deformable objects too.
Third, we saw it as a challenge. Doing laundry was one of the tasks scientists couldn’t get robots to do. Before we started working on this project, no comprehensive success story had been reported for the complete end-to-end task of reliably picking up a laundry item and folding it.
SmartPlanet: What has changed in the field of robotics?
PA: Back in the early days of artificial intelligence research, in the 1960s, one of the goals was to build a robot that can serve in our households. One of the most notable attempts was Shakey. It wasn’t easy. Artificial intelligence researchers began to focus on figuring out logical reasoning, computer vision, motion planning, machine learning and control. Switching gears led to significant progress in each of these subfields. This progress has, of course, also been driving by advances in electrical engineering, with more accurate sensors and more computational abilities.
SmartPlanet: What are some other cool robots that are going to help us around the house?
PA: It’s difficult to predict which robots will eventually make it into our houses. It’s possible one of the following research personal robotic platforms: Willow Garage (PR2), Meka Robotics, HRP-3 and Justin.
SmartPlanet: If you could program a robot to do anything, what would you make it do?
PA: We aren’t trying to program a robot for a specific task, but are trying building robots so they have the ability to perform a wide variety of tasks.
SmartPlanet: How do you think we should treat robots?
PA: One of the main things that always strikes me is that thinking of robots as “beings” is not that likely to happen when you actually program the robot, and know it is essentially executing what you coded it to do.
Related on SmartPlanet: