Dog training methods are now being used to teach robots.
(No, it’s not April 1st and no, I couldn’t believe it either, but it’s true – I can give you the journal reference if you don’t believe me!)
Robots are so sophisticated now, they are able to learn different tasks. But most of their training to date has been done through trial and error and it takes a long time. Then a dog owning scientist, a Dr Hundt (wonderfully suitable name!) had the bright idea of using positive reinforcement. Just like we use when teaching a dog – rewarding the behaviour you want.
The aim was to teach Spot the robot (yes, really!) to stack blocks. He needed to focus on the constructive actions to take to achieve the aim. Previously, teaching this task to a robot took weeks. So would the dog training approach work?
Now obviously Spot would not appreciate bits of kibble and cheese or chicken would make a nasty mess of its expensive electronic insides, so the scientists devised a points reward system – apparently robots love earning electronic points.
To stack blocks, Spot the robot needed to learn how to focus on constructive actions. As Spot explored the blocks, it quickly learned that correct behaviours for stacking earned high points, but incorrect ones earned nothing. Reach out but don’t grasp a block? No points. Knock over a stack? Definitely no points. Spot earned the most points by placing the last block on top of a four-block stack.
“The robot wants the higher score,” Hundt said. ” At the start the robot has no idea what it’s doing but it will get better and better with each practice. It quickly learns the right behaviour to get the best reward. It never gives up and keeps trying to stack. In fact, it used to take a month of practice for the robot to achieve 100% accuracy. We were able to do it in two days.”
Positive reinforcement not only worked to help the robot teach itself to stack blocks, with the point system the robot just as quickly learned several other tasks. The ability to learn from mistakes in all types of situations is critical for designing a robot that could adapt to new environments.
The team imagines these findings could help train household robots to do laundry and wash dishes – tasks that could be popular on the open market and help older people live independently. It could also help design improved self-driving cars.
“Our goal is to eventually develop robots that can do complex tasks in the real world — like product assembly, caring for the elderly and surgery,” Hager said. “We don’t currently know how to programme tasks like that — the world is too complex. But work like this shows us that there is promise that robots can learn to accomplish such real-world tasks in a safe and efficient way,” Hundt said.
Fascinating! Age old principles are robust. As you know, training your dog using positive reinforcement works too – your dog will keep trying and eventually succeed, just like Spot.
Keep training and stay safe,
Carol