Jibo
For many families, the tablet has become the central, shared computing device in the home. It’s a hub for learning, for entertainment, and for staying connected. But what if your tablet was even more interactive? What if it woke up when you can home, recognized your face, and suggested a couple of things you might want for dinner? What if, when asked a spoken question, it could tailor its answer directly to you, instead of just offering a blanket response?
A new device called Jibo can do these things, and it could mark the next step in group computer interaction in the home. But Jibo isn’t a tablet at all: It’s a robot.
Specifically, Jibo is a social robot. You talk to it, ask it questions, make requests. It talks back, provides answers, and takes care of grunt work like setting reminders or scouring the web. It’s meant to act as a helper and a partner in a variety of household experiences, much like a physical embodiment of Siri, Google Now, or any of the voice-activated concierge services available on our smartphones or tablets.
But unlike those handheld touchscreen devices, Jibo tries to act like more of a participant than a tool, as if it’s a part of the family. It has a big round head, and a face that “looks” around the room. The foot-tall, bulbous body can rotate to address the person speaking. It even leans a bit when it turns to face you, as though it’s listening more intently.
Jibo is only a prototype right now. The team behind it, headed by founder Cynthia Breazeal, who is also director of MIT Media Lab’s Personal Robots Group, hopes to bring it to market in time for the 2015 holiday season. Curious early adopters can join the crowdfunding campaign that begins today. The pre-sale price tag is $500 for early backers, and $600 for a developer kit. That’s a little more than the cost of a good tablet. And Brezeal is clear about how Jibo is designed to perform the same types of interactions families currently use tablets for, but to do so with a physical presence that fits into human lives in a more natural way than just another touchscreen.
Like a tablet, Jibo can take photos and videos. It can pull up information from the web or an app, it can act as a teleconferencing device, and it can be used to queue up books or videos. Using a mixture of facial and voice recognition (as well as an iOS and Android app), it personalizes these experiences for you. You can ask Jibo to order your favorite take-out Chinese meal after arriving home from a late night at work. Or tell it to display an e-book on its face-screen, turning a storybook into an interactive, theatrical experience for you and your child. It can recognize and greet you when you get home, or remind you to make an important phone call in between the day’s errands.
“We need technology to transcend the world of information into a more humanized realm,” Breazeal told WIRED. The connected home of the future shouldn’t feel cold and computerized, operated with Star Trek-like voice commands, she says. It should be warm and personal, interacting with us on an emotional level in addition to being able to perform useful tasks.
And thanks to the mobile computing revolution, for the first time, sensors and processors are small, efficient, and cheap enough for something like this to take the form of a robot that’s both priced and sized reasonably enough for consumers.
“Something like this is a nice bridge between devices and tablets and robots that we imagine in science fiction,” Breazeal says.
How It Works
One of Jibo’s key features is human and facial recognition. Using a stereo camera system, it can distinguish people from their background surroundings so it knows when there’s a person in the room. In particular, it can recognize faces, so it knows which human it’s talking to. When development is complete, Jibo will also be able to recognize facial expressions so it can guess your mood and cater its interactions to your current state of mind.
On-board hardware includes a 360 degree mic array so the robot can perform sound isolation, identifying when it’s being spoken to even if the person talking is not right next to it. Dual speakers supply its voice and other audio. Wi-Fi and Bluetooth radios keep it connected. A quad-core ARM processor act as the brains. On its face is a circular LCD touchscreen, and its plastic “skin” is also touch-responsive. A 3-axis motor system allows the top section to spin all the way around on the base. While it’s meant to stay plugged in the majority of the time, it does include a battery so you can move it around the house for short periods.
Interface and Design
Though Jibo is still a prototype, Breazeal’s team developed a demo to show what the robot will eventually be fully capable of in terms of looks and behavior. The appearance is close to final. Jibo actually looks a lot like Eve from the movie Wall-E, at least in the prototype I saw. The body is shiny, circular and white. The head is spherical, though a chunk is cleanly sliced out of it so a flat LCD display can act as its face. “For a while, we were excited about curved displays, but we realized that the technology wouldn’t be ready and robust enough,” Breazeal says.
Jibo turns to look at you when you talk to it.
The head and body can both rotate 360 degrees, so the robot can rotate to look at whoever is speaking to it, or just swivel and twist animatedly as it responds and interacts with you (kind of reminiscent of the Keepon robot).
As for the onscreen user interface, Breazeal added a character animator to the team to handle that task. Instead of some sort of app or list menu as an interface, or a human-like face, Jibo’s screen displays a simple, white sphere. This ball can morph into other graphical elements: a clock, an illustration of the weather, a heart, a smile. It’s designed to be dynamic and easy to read from across the room. It comes across as friendly, familiar, and expressive, all without being too cute, or verging anywhere near the uncanny valley. It’s technology humanized, but not necessarily in humanoid form.
While the prototype is expectedly rough around the edges—the LCD is low-res, and the robot’s movements are sometimes too abrupt and swift to seem natural—the potential is clear.
Jibo takes what we’ve learned from smartphone and tablet experiences, specifically from voice interactions in systems like Google Now, and builds on it. It does much of what the software on your devices can already do—learn your preferences, predict your needs—but it does everything with more personality. And whether Jibo succeeds or fails depends a lot on how that personality jibes with the humans who have to live with it.
No comments:
Post a Comment