UPenn's DORA telepresence robot
University of Pennsylvania
SHARE

Like its namesake, DORA was born to explore. Specifically, the robot—which was built by a team of students at the University of Pennsylvania—is designed to be a kind of exploration surrogate, able to move its head with the same speed and flexibility as the human seeing through its eyes. DORA’s movements are mapped to an Oculus Rift virtual reality headset, so when the goggle-wearer swivels left, the robot follows suit. It provides something seemingly unprecedented in robotics: telepresence from the neck up.

In fact, that’s pretty much all DORA, or Dextrous Observational Roving Automaton, can do. It’s a robotic bust mounted on a four-wheeled, battery-powered cart, so it can navigate itself through environments. But mobility isn’t the focus of this research project. “This is about recreating the experience of being in another place,” says John Nappo, an undergraduate roboticist at UPenn. “We thought, How do we capture this in as realistic a way as possible?”

DORA’s primary mission is to breath life into consumer-grade telepresence robots, a class of machines that are little more than webcams on wheels. Companies like Double Robotics and iRobot offer telepresence models that upgrade the telecommuting experience—employees can remotely swing by a cubicle or conference room, rather than simply joining via speakerphone. These are not immersive systems, however. The robots’ operators watch video from a single camera, and often have to reposition their surrogates entire awkward body in order to see who in the room just piped up, or to pick up the subtle, nonverbal social cues that enrich human interaction.

Of course, you don’t see most telepresence bots outside of office or work environments, because they’re only barely useful in the highly structured world of corridors and cubicles. If a physically disabled operator wanted to check out a museum exhibit, for example, today’s robots do a worse job of transporting the user than a smartphone-based Skype connection. This is the largely boring, and unsophisticated state of consumer telepresence.

DORA, at least on paper, is different. Never mind the robot’s mobility. It’s the bust that matters, and the machine’s makers imagine DORA-like systems planted in places that deserve to be viewed as if you actually there. Stick one in the Sistine Chapel, and remote visitors could get some sense of the work’s grandeur, and without the distorted, stitched-together view that Google’s Street View cameras (and similar systems) provide. Along with perfectly mirroring the way a human moves his or her head while perceiving the world, the robot’s dual cameras mimick our stereoscopic vision. The user sees in 3D, and not just the clumsy layers of flat, paper doll-like images that Hollywood often passes off as 3D. “You can see the rounded edges and curvatures of cables, for example,” says Nappo.

DORA, in other words, could be useful for applications other than sightseeing. Tasks that require fine, precise perception and manipulation could benefit from a swiveling, stereoscopic-visioned robot head. “We didn’t built the prototype to handle this, but we can see it being used in first response,” says Peter Zachares, a mechanical engineer at UPenn. From bomb disposal to emergency triage, Zachares believes DORA could help “wherever the user needs to be up close and personal.”

In fact, it’s hard to come up with a kind of teleoperated robot that wouldn’t benefit from a fast-swiveling robotic head. From armed drones and military ground bots to the walking, humanoid disaster responders in DARPA’s upcoming Robotics Challenge, the ability to at least occasionally strap on an Oculus Rift and take a remote look around could be more useful than an entire suite of sensors.

Nappo, Zachares and two fellow students have been working on DORA for the past year, and it’s been selected as one of 20 robots that will compete in next week’s Intel-Cornell Cup in Cape Canaveral, Florida. The top prize at the student research competition is $10,000. But whether or not the robot takes home any cash, the event will give DORA its biggest stage yet. With enough funding, Nappo believes that DORA could be commercialized in around two years, with units priced at roughly $2000 each. That’s cheap enough for robotic heads to actually start showing up at tourist attractions, in conference rooms, and on other bots. If the Oculus Rift becomes the game-changer that so many analysts expect it to be, VR-controlled robots won’t be far behind.