Yesterday, OhmniLabs (a robotics startup based in Santa Clara, Calif.) launched a new consumer telepresence robot on Indiegogo called Ohmni. With a lightweight design, integrated screen, and easy to use software, Ohmni is designed to be the most accessible (and affordable) mobile telepresence robot yet.
Ohmni is heavily focused on ease of use, which is always a good idea when developing a robot for consumers. It arrives completely assembled (folded in half), and setup consists of unfolding it, meaning that you can send it to a relative with minimal computer experience and not have to worry. The display, audio, and video hardware are all integrated, so there’s nothing to mess with there, either.
The robot itself is 4’ 8" tall, and weighs just over 18 pounds. It runs Android 6.0 on an Intel X5 processor, with 64 GB of on-board storage and 4 GB of RAM. The lithium battery will keep the Ohmni moving and chatting for 5 hours, and the included dock will give you an hour of life for every hour of charging. Ohmni does include an autodocking feature with minimal autonomy: as long as the cameras can see the fiducial on the dock, the robot can orient and dock itself.
One feature you’ll find on Ohmni that distinguishes it from other telepresence platforms is a tilting neck. Not pan and tilt, because it’s not designed to let you look around, rather it’s just a single extra degree of freedom that allows Ohmni to look up, down, and nod. You control this manually, but I can imagine how this simple change would go a long way towards making the robot seem more connected to the person remotely inhabiting it.
For more details on Ohmni, we spoke with Jared Go, Ohmnilabs co-founder and CTO:
IEEE Spectrum: How do you convince people that a telepresence robot is worth investing in when the cellphones and tablets they already own come with video chatting capability?
Jared Go: One key part is helping them realize that they’re just taking for granted what it means to stay in touch by phone or video chat. Almost everyone we talk to tells us they miss the feeling of being there in person, even with all these video chat apps and hardware surrounding us. So there is definitely a value gap there and our research and user testing early on focused on that key question—how to use robotics can help solve that problem.
We know now that the key lies in the holistic, differentiated experience. It’s not about having a screen, or remote control wheels. It’s about what this enables. We studied behavior as we let testers try Ohmni with their own family and the results were really awesome. Almost the very instant they tried it, the way they interacted completely changed.
For example, we found conversations would immediately move to the most natural and comfortable place, rather than say sitting in the study or on a bar stool. The other side most often said they felt way freer, not forced to constantly aim a device at them, or sit in one spot. Because of this, conversations got longer and instead of just recapping what happened last week, people would spend an hour or more just hanging out around different activities, like watching sports on TV. This is stuff they’d never do with FaceTime or Skype.
Double and Beam have designed their products around the corporate use case from day 1, which has forced them down a particular road very different from us. We’ve focused on home from the very beginning, which allows us to provide a much better suited experience. We’ve rethought the entire process. The tablet on our robot is not removable and is completely integrated and set up ahead of time. When you order, we can even pre-program WiFi credentials for you so that if you ship it to your parents, they just unbox, unfold, and turn it on. That’s it! We’ve had people 65+ set it up with high level verbal instructions only. That’s super important for the consumer market.
The moving neck along with our MotionMap technology is also key to creating biggest difference in experience. Driving around Beam and Double and interacting feels a lot like walking around with a neck brace on, which results in unnatural interaction. We studied this and notice in a lot of cases people will bend down or adjust their pose to look flat into screen, because the screen and camera can’t orient to look straight at them. People say that it’s really the moving neck that makes the product feel really human. The human brain is hardwired to sense attention so seeing what the remote user is exactly looking at has a tremendous impact. Being able to emote and nod too really gives the extra oomph to that feeling.
Why should roboticists be excited about Ohmni? What kind of interesting technology (software, hardware, or both) does it use?
Two main things. We’re focused on optimized the telepresence experience first, but we’ve already started building out Ohmni as an affordable, human-scale open platform. We have a cloud-based JS framework we call OhmniAPI. Our goal here was to make programming Ohmni as easy as developing web pages. You basically develop a rich HTML5 app that can run display whatever complex media or network logic on the screen and also uses privileged JS calls to have full access to motors, lights, text-to-speech, and so on. And you deploy it to any bot with one click from the cloud. We’ll be extending the capabilities of the framework over time to add higher-level navigation and image/audio recognition so that writing high level tasks will be possible without a Robotics PhD.
We’re also developing an in-call developer console so you can run code on the bot while you’re in the telepresence call, which we think is a really exciting way to do robot development! We’ll release more details on this later on in the year.
Ohmni’s also built on open standards so it gives devs a lot of flexibility to hack on the hardware. We’ve left extra hardware ports where you can attach extra servos and sensors. The core of Ohmni’s sensors and peripherals is a USB root hub so you can easily plug in and control all the hardware from a beefy laptop or anything else too. We’ll be publishing the packet protocol for controlling motors as well.
How do you plan to improve Ohmni through software updates?
We’re always iterating on things like autodocking precision, MotionMap precision. Continually improving the driving UI. We’re planning to roll out features over time like hotword detection and speech control (i.e. integrating with the Alexa API). We’ve definitely got plans for more autonomy and more assisted driving and will talk more about that later this year. We’re hoping developers do crazy and fun things with Ohmni like try and do some interesting HCI applications or even things like get a whole set of Ohmni bots to sync together and perform songs from Hamilton*.
You can pledge US $1460 (shipping included) for an Ohmni on Indiegogo right now, which is $500 less than the robot will eventually sell for. This is likely to be the biggest issue with Ohmni—even though it’s relatively affordable for a mobile telepresence robot, its price will make it a luxury item for most people, and it could be hard to justify.
We certainly agree that telepresence robots have a surprising amount of value, and that it’s something it’s best to experience to really understand how it’s different from (and better than) a regular video chat: the sense of embodiment you get is powerful, and the ability to have a physical presence somewhere else makes interacting with people remotely a much more natural experience. There’s a lot to like about Ohmni, and the platform in general seems to have potential. While it may not represent a significant advance in terms of mobile telepresence technology, it’s a compelling mix of features and affordability that seems to have potential.
[ Indiegogo ]
* I had to Google this, of course, and while I wasn’t able to immediately find any robots singing songs from Hamilton, I did find these YouTube videos of FRC teams performing some entertaining parodies: