Wagging Tails Help Robots Communicate With Humans


Giving a Roomba a tail makes it easy for humans to understand its “feelings”

Photo: University of Manitoba HCI Lab
“Wait, is that a tail on that Roomba??”

I have no idea what my Roomba is doing most of the time when it runs. It’s vacuuming, I know that, but sometimes it just sits there for a little bit, or slowly swivels back and forth, or does something else that doesn’t seem (strictly speaking) vacuuming related. This isn’t as much of a problem for Roombas specifically, but for robotics in general, it can be: If robots are bad at communicating what’s going on with them, it’ll be harder for people to accept them in our daily lives.

One thing that lets humans instantly grasp the abstract internal state of other humans is we look at each other’s faces. Now, as you can imagine, giving robots human faces can lead to other problems. The good news is we’re also hardwired to perform this intuitive abstract internal state reading trick on some other expressive living things, like dogs: When we look at a dog’s tail, we get an indication of whether it’s happy or not. It turns out that we can do the same for robots, as long as you can give them a tail.

Roomba uses tail to communicate with humans

Image: University of Manitoba HCI Lab
iRobot Create equipped with an actuated, fluffy tail that it can wag like a dog.

A few years ago, University of Manitoba undergraduate student Ashish Singh and professor James E. Young decided to investigate whether people could accurately interpret the “feelings” of a Roomba with an actuated, fluffy tail that it could wag like a dog. The Roomba doesn’t have feelings, of course, but acting “happy” could mean that all systems are okay, while “sad” could communicate a problem and “tired” could mean a low battery state. In results published in 2013, they found out that it works:


About Author

Leave A Reply