While fears of rebellion have underlined robot intercommunication since Skynet, that hasn’t stopped scientists from developing an open-source world wide web called RoboEarth for robots to learn from each other–the first step toward interdependent robot learning.
A robo-specific Internet has obvious benefits: if one robot has been in a room, that mapping info can be instantly shared with others. Ditto with a person’s medical history or even medical procedures. In short, the concept of “cloud robotics” means that a procedure would only need to be taught once for any robot hooked to the network to replicate it. As the BBC reports, the eventual goal is to create a cloud-based database that would operate as a common brain for machines–and best of all, RoboEarth’s software components are open source.
And like a smart home, some of the heavy computation can be offloaded to other sources on the network, reducing weight, price, and complexity of robot brains while extending battery life. Further, household robots with different abilities could work as a team using their different mechanical capabilities, whether that means sending the Roomba under the couch to bump out a submerged toy or using multiple robots to lift heavy furniture.
For now, research scientists behind the RoboEarth project are testing a quartet of robots at Eindhoven University. Using a mock hospital environment, the scientists are testing navigation and simple tasks like serving drinks to patients.
RoboEarth began in 2009 with funding from the European Union and developed by scientists from Philips and five universities, including Eindhoven University.
Like any network, RoboEarth would theoretically allow overseers to monitor many user robots at once, which James Barrat, author of the futurist book Our Final Invention, believes would provide a bottleneck to prevent robot malfeasance. In the long term, however, Barrat suggests that RoboEarth might allow robots too much opportunity to learn from each other–and start improving their own software for their digitally nefarious ends.