Everyone can agree that robots will become more commonplace in our homes, public spaces, and work environments. As that happens, they’ll need to know how to operate harmoniously with humans. That doesn’t just mean that your Roomba will be able to vacuum your apartment without sucking up your cat, it also means they will need to understand human social norms. That’s why researchers from the Yale Social Robotics Lab have created OwnageBot, which understands the concept of ownership.
Through a combination of experience and explicit rules, humans learn to intuitively understand ownership. You know not to eat food that isn’t yours from the office refrigerator, because it’s safe to infer that it belongs to someone else and they’ll be hangry if you take it. You know not to take your neighbor’s car, because that’s felony grand theft auto. OwnageBot understands the same concepts through a machine learning system that mixes explicit rule sets with experience-based inferences.
OwnageBot, which appears to be built on the Rethink Robotics Baxter robot platform, was designed specifically to exhibit respect for object ownership in order to seamlessly integrate into humancentric environments. Say you’re working in an auto garage alongside a mechanic-bot, for example. You own your tools, but your robot coworker may need to borrow a screwdriver. A human would know to return the screwdriver, because it belongs to you. That’s not something a robot just knows, which is why the OwnageBot research is important — so you get your screwdriver back.