One method for doing this is the TellMeDave project from Cornell University researchers. They have developed an online game that allows volunteers to train a robot to perform various tasks and respond to normal language.
Each iteration of a task is stored in an extensive central brain called RoboBrain that is then made available to a wide range of robots.
Robots teaching robots
A more interesting trend however is where robots begin to teach one another. I’ve written previously about various projects that aim to achieve this, such as the MIT lab for collaborative waiters, the European CARLOS project that is putting collaborative robots to work in shipyards and the improv jazz robot being developed by the US Defense Department.
A Brown University project, codenamed Baxter, is building on both the human driven and robot driven approaches to learning. The aim is to better understand how robots can collaborate effectively with one another.
“It’s pointing in an interesting direction,” the team say. “When you put a robot in a new situation—and in the real world it happens in every room the robot goes into—you somehow want that same robot to engage in autonomous behaviors.”
It’s undoubtedly the case that robots will operate in a collaborative way in the very near future, so extending our knowledge of how best to achieve this will be a vital step. This is particularly challenging when the environments in which the robots operate are fundamentally different.
The ultimate goal is to develop robots that are capable of figuring out for themselves how best to translate the information they’re receiving based upon a real time analysis of the robot sending them the information and their various similarities.
Suffice to say, that is still a little way off, but there are undoubtedly solid steps being made towards that end goal, and the capabilities of robots will rise incredibly far should it be reached.