In May of last year, Misty Robotics raised US $11.5 million and spun itself out of parent company Sphero. From the sound of things, Sphero cofounder Ian Bernstein had been working on an idea for a new kind of robot inside Sphero for a while, and Misty is the manifestation of that. According to the spin-out press release:
Misty Robotics’ vision is to put a personal robot in every home and office. These robots will be seen and treated as our friends, our teammates, and a part of our families—performing helpful tasks, providing safety, and interacting with humans in entertaining and friendly ways that have only been seen before in science fiction. Misty will begin its mission with the release of its first robot and a collaborative ecosystem for robotic development.
That’s quite a vision, especially in the context of the sky-high expectations that Jibo isn’t (yet) living up to, and that Kuri may struggle to meet. Of course, this is Misty’s long-term vision, and they fully expect it to take years to get there. In the short term, they’re starting with a platform for developers. Not roboticists, but software developers, and people without much in the way of experience programming robots.
Today, Misty is announcing the Misty I Developer Edition prototype, designed to “provide developers and inventors with everything they need to create the ‘killer app’ for personal robots.”
The press release about Misty I doesn’t contain much in the way of detail, but here’s as much as they’re giving us:
The Misty I Developer Edition robot will begin shipping next month and is priced at $1,499.
Image: Misty Robotics
Enwall: [Misty I, the black one] is a hand-built version. The [white version, Misty II] is one we’re working on with our manufacturer in China. It’s mass produced, with a great industrial design. It’ll be available later in the year. What we decided was, we really wanted to start working with a handful of “superdreamers.” We designed a hand-manufactured robot that we’re going to release several dozen of, running an early iteration of the same software that the production robot will run.
Bernstein: Misty I is mobile and 18 inches [45.8 cm] tall. It has track drive, with an active depth sensor that we use for mapping and localization. We’re working with a company called Occipital: They make a sensor called the Structure Sensor. We also have an HD camera; we can do object recognition, face recognition, face tracking, things like that. We’re using Qualcomm beamforming microphones to detect sound and directionally understand where it’s coming from as well. There are high quality speakers, a display, and time-of-flight sensors for additional object sensing.
We’re really targeting developers, by creating a robot that’s open and super easy to program, and basically solving a lot of the hard problems in robotics and allowing developers to do more interesting things with robots.
IEEE Spectrum: So Misty is designed at least initially to be a developer platform rather than a robot that consumers will buy, and you’re hoping that those developers will create a lot of useful content for the robot?
Bernstein: Exactly. It’s really hard to build all the applications that a consumer would want to allow the robot to do useful things in the home, so our idea is to target developers so that we, along with a community of people, can develop a lot of really cool stuff.
Enwall: It’s important to note that both the developer robot and the production robot are targeted at programmers. We’re pretty confident that 90 percent or more of the people who are going to want to buy the robot will be programmers, because that’s who it’s made for. We’re fairly big believers that, especially in North America, consumers aren’t ready for, aren’t interested in, aren’t demanding robots in their home. I think one of the reasons they’re not is one of the first things they ask is, “Well, what does it do?” And the answer that you really want to give is, “It does these hundred things.” No robot company on the planet right now has the resources to do that, but if you release it to tens of thousands of developers, all of a sudden the robot can do a hundred things.
IEEE Spectrum: How will you convince developers to buy a robot and spend their time developing apps for it?
Enwall: The answer to that is, “Because it’s a robot.” We’ve been talking to a lot of developers, and the questions that we ask them is, “If you can program a robot to do something meaningful in 30 minutes, would you be interested?” And almost everybody says, “A real robot?” And you say, “Yeah.” And they say, “I’m definitely interested.” So our whole hypothesis is that there are tens of thousands of latent dreamers, your Web developers and your mobile programmers, and our goal is to unleash them with Misty in the same way that the personal computer and the Web browser unleashed millions of programmer dreamers—that’s the way both those revolutions were born. Millions of dreamers went out and pursued their dreams, and then they invented things that all of us wanted in the office and the home.
IEEE Spectrum: I’m not sure that this has been a successful model with robots before; how is Misty going to be different?
Enwall: We believe that the things that have come in the past have been too expensive, too clunky, not capable enough, not well designed, not easy enough to program… We believe we’ve gotten the combination of price, capability, design, and ease of programming at an optimal mix so that you’re average Web developer will look at this and say, "Wow, that is affordable and super powerful. I want to do something with it."
Bernstein: Part of this is taking our experience and the lessons we’ve learned from shipping several million robots with Sphero—we know how to build a robot that’s super easy to program, and it’s in well over ten thousand schools now, worldwide.
Photo: Misty Robotics
IEEE Spectrum: So it’s not that you’re trying to sell developers on generating content and becoming part of an ecosystem, it’s that you’re trying to sell developers on having their own robot that they can get it to do something cool?
Enwall: Yeah. And they might do something cool for their business, or even to create a business.
IEEE Spectrum: What are some specific examples or categories of the kinds of things that you’re hoping that developers will come up with?
Enwall: All kinds of things. We’ve got dozens and dozens of lists. Detect a pet and perform an action. Detect a family member or an office coworker and perform an action. Read social media or news or weather or bedtime stories. Perform an action based on hardware extensibility like controlling a smart home. Post a photo on social media. Record video when motion is detected. One that came up yesterday was showing what a Wi-Fi signal is at any particular part of a home or office.
IEEE Spectrum: How will you make Misty easy to program for a developer who might not have any experience with the sorts of things that make mobile robots uniquely capable?
Bernstein: A lot of the functionality will be abstracted at a high level. Instead of having to deal with SLAM, you could just tell the robot to go to a specific room. As a developer, you won’t have to know how the map is generated, or even what SLAM is. And then of course we’ll provide the functionality for someone who’s more experienced so that they can dig down to more complex stuff. We’re trying to balance diving developers enough tools so that it’s not too complicated, but there’s enough there that you can actually do some useful stuff.
Enwall: Seven years of building robots for kids [at Sphero] is a big differentiator for us. It’s taught Ian and his team members how to make a really enjoyable, awesome robot that’s easy to program, and how to take all of that complexity and make it simple.
IEEE Spectrum: Your press release says that your vision is to have personal robots “as our friends, our teammates, and a part of our families.” How will you manage expectations for what Misty is capable of?
Bernstein: I think part of it is introducing that type of stuff when the time is right, when enough skills have been developed for the robot. Misty I is targeted towards developers, and we’re not trying to come out of the gate with a robot that has personality and creates strong personal connections. That part will come later.
Enwall: It’s a journey. We’re certainly not going to release an iteration in 2018 that somehow magically transports us 15 or 20 years into the future of AI and personality. We’re not going to consumers and saying, “Here’s your social buddy.” We’re going to programmers with tools that they can use to help create some level of human connection through the experience of a skill. And again, we’re letting the world of millions of dreamers and inventors take those basic tool sets for connecting with humans and expand them and explore and adapt so that over time, yes, these robots become like your colleagues or your friends or your family members. It’s at least 10 years for that level of experience to have a shot at happening.
IEEE Spectrum: It sounds like it will be a long process, an intentionally long process, to get Misty to the point at which you might call it a consumer robot.
Bernstein: Yes, exactly. Like with the computer industry, it took four or five or six years to get to that stage—it took years to get to VisiCalc, which was the first killer application for computers. And that’s really when computers started to take off like crazy. We don’t know how long it will take us to get to our VisiCalc.
IEEE Spectrum: Why is now the right time to put a robot like Misty in the hands of software developers?
Bernstein: We looked at it from both a user readiness standpoint and a technology standpoint. Technologies are starting to get good enough and cheap enough that we can actually create this robot.
Enwall: It’s a price-performance curve, where, finally, we can deliver for about the price of a decent laptop, a bunch of great capability.
Enwall: I think both of them are wonderful for the robotics industry. There are awesome, talented people working at both companies. Jibo has done a great job on the social, personality side, but it doesn’t move around, and it can’t act with independent autonomy, which is a pretty severe limitation in our minds for all of those dreamers who would love to program a robot. And then Kuri, I think they’ve done a great job, but they don’t have something for programmers. It’s geared for some specific use cases, targeted at the consumer.
We don’t think consumers are ready, en masse, for these kinds of advanced capabilities. The expectation of a regular consumer, about what something that is enabled by voice and understands language and moves around can and should do—those expectations are so high. We don’t think consumers are ready for robots; we think programmers are, because it’s going to take three to five years of refining and working and building real, meaningful experiences that will meet the expectations of regular consumers.
Photo: Misty Robotics
IEEE Spectrum: If I’m not a developer, but I like the idea of a useful personal robot like Misty, when can I get excited about it?
Enwall: I think a couple years from now. Misty I is accessible through Blockly, which is a graphical programming tool, so there is a middle ground for someone who wants to experiment with what an independent robot could do for them. But I think really the basket of skills that most consumers will be interested in will take three or four years to develop.
Bernstein: It’s really hard to say. I’m excited to be using the product myself, because I want to build things that I think are cool. Eventually we’ll get enough skills that a bunch of those will be cool to other people as well, and they’ll just want to buy the robot and download skills that other people have written. We don’t know—it could take six months, it could take four years, but we’ve built the company for that.
Enwall: Our whole hypothesis is that there’s a latent world of programmers who, if you gave them a really great robot to program that was affordable, would go nuts. We have to execute on that—we have to create a great robot that’s affordable and easy to program, and if we can execute, we believe that that latent desire sits inside a lot of programmers.
We certainly believe that Misty is capable of making a great robot. The question now is, do people want a developer platform like Misty? Will developers find this vision compelling, and will they decide that it’s worth it to invest a substantial amount of money in a platform that won’t really do anything until they also invest their time and energy into making it work?
Misty will also have to compete against existing smart home hardware. Such hardware may not be programmable (save very simplistically through IFTTT-like services), but it’s cheap, readily available, and does what you want right out of the box. Besides sophisticated programmability, Misty has another advantage over traditional home automation: mobility, combined with its ability to create maps. It’s likely that the most compelling use cases of Misty will take advantage of this.
So far, robots designed for the home have been forced to lean heavily on the social aspect—that’s what differentiates them from a security camera, or an Amazon Echo, or a Roomba. It’s not yet clear whether this is turning out to be a successful approach, with reviews of Jibo being decidedly mixed. Without that social component, at least initially, Misty could justify itself by allowing people to create functionality that is either versatile enough to take the place of a bunch of other home automation services at once, or unique enough (think “killer app”) that it’ll be worth it to get a Misty for a few specific reasons.
People have been trying to find an achievable killer app (or apps) for household robots for a long time. (Bill Gates wrote his famous “A Robot in Every Home” article in 2006!) So it’s a bit of a risk for Misty to be counting so heavily on a developer community that doesn’t exist yet to come up with a whole bunch of amazing ideas that might, in the future, make Misty II worth it for nondevelopers to buy. To their credit, Misty is already investing in that community, having brought on Ben Edwards, founder of SmartHome, as head of community.
The folks at Misty know all this, of course, and they’re all industry veterans with the experience to potentially make Misty work. One of the things that makes me the most optimistic about Misty is that the company seems to have a realistic perspective on how long this might take—it will definitely be a few years, and it may be many years, before Misty will appeal to someone who doesn’t identify as a developer or hobbyist. And over that time, Misty will be facing competition, to some extent, from robots like Jibo, Kuri, and Buddy, and possibly also from companies like iRobot and Neato, which are beginning to leverage the mapping and smart home integration capabilities of their vacuums.
We’re as anxious as anyone for home robots to become useful, but we’ve been hearing about how close we are for long enough now that cautious optimism about new products like Misty is about as much of a commitment as we’re comfortable making. Having said that, we are excited to see what developers start doing with Misty, once it’s released next month. And if you’re already sold, you can apply for the Misty developer program at the link below.
[ Misty Robotics ]