As we expected, there were no major robot announcements at CES, because that’s not really what happens at CES. That’s not to say that there were no cool robots at CES, because there were. One of them was a personal robot Kickstarter project from New York City start-up Robotbase.
It caught our eye because it looks like it’s built from a TurtleBot foundation, which is a great way to start, but the rest of the robot looks really nice as well. Robotbase has ambitious plans for it, especially in terms of software. If it delivers on its promises, this could become a really compelling platform for consumers and developers—and we might even start believing that this whole personal robot thing is finally happening.
Here’s the Kickstarter video. It’s appropriate to be a little bit skeptical of it, because I usually feel like the slickness of a video is inversely proportional to my initial confidence in what the end product will be.
They are promising a lot here—a host of amazing applications, a flawless user interface (UI), superb natural language processing, smooth Scarlett Johansson-like voice synthesis, and more. In terms of the hardware alone, you’re looking at a Turtlebot 2 with a whooole bunch of extra stuff going on. Seems pretty impressive, especially for the Kickstarter price of US $1,200.
We asked Robotbase founder and CEO Duy Huynh to highlight how the Personal Robot is special:
Our robot is a superset of Turtlebot 2. It can do everything that Turtlebot 2 does, plus a lot more. If you look at Turtlebot 2, it's a depth camera on top of a mobile base. We have the same Kobuki base and the same depth camera, so anything you can do with Turtlebot, you can do with our robot. But that's only a very small part of our robot. We have a lot more:
1. Built in Z-wave, Zigbee, Bluetooth, and Wifi. Our robot can interface with almost all of the connected devices on the market today, and for the first time, developers and researchers can begin to work with the data from connected devices.
2. Sensors. We include various sensors such as temperature, humidity, CO2, etc. This is extremely helpful for developers and researchers who want to work with environmental data.
3. Software Stack: ROS is great. We added more. We'll ship APIs around facial recognition, object recognition, natural language processing, speech recognition, etc. Developers and researchers can leverage these AI APIs right away, without building them from scratch.
4. A UI. This was one of the main problems we had with Turtlebot: there was no UI. Not only do we ship a UI, we ship a touch UI. This opens up a huge range of applications, in both consumer use cases and enterprise use cases. For the first time, developers and researchers can build applications with practical use cases outside out the lab.
5. Form factor. Turtlebot is less than 2 ft tall. Our robot is more than 4 ft tall. There are a few advantages with that. First the camera is at a much higher position, so it looks at things pretty much the same way we human do. Second, interaction with the robot is just much more natural.
The problem with TurtleBots and robots like them is that while they can do lots of things, they don’t actually do anything unless you program them to do something, which is hard, especially with a small mobile robot that doesn’t come out of the box with an easy way to interact with much. The additions that Robotbase has made to their Personal Robot means that you have a chance of easily getting them to do something useful.
And there’s more news on that front, as Robotbase is announcing that their robot will also be available as a developer/research version that will optionally ship with an Android operating system pre-installed, opening up an entirely new app ecosystem.
For $1,495, you can get a Kobuki mobile base, a 3D depth sensor (an Asus Xtion or something similar), an Inforce 6410 single-board-computer (powered by a Qualcomm Snapdragon 600 processor), nice Dynamixel servos in the robot’s neck, and all the rest of the sensors and wireless connectivity and stuff that comes on the full Personal Robot version. Plus, you’ll get to choose whether to get Ubuntu/ROS or Android as an OS pre-installed. Compared to a stock assembled TurtleBot 2 at $2,000, it’s a pretty impressive list of features.
We’re very much liking the hardware and the philosophy here, so from our perspective, the tricky part is going to be the software, and (relatedly) managing expectations. Robots are hard, and consumer robots in unstructured home environments are ridiculously hard.
As always with Kickstarter, you’re buying into a dream, and the dream here is going to be a real challenge to fulfill. After talking with Duy, we’re optimistic about this project, because if nothing else, you’re getting a lot of solid hardware and capability for your money. If all of the software components and apps work as advertised, even better.
[ Kickstarter ]
Evan Ackerman is a senior editor at IEEE Spectrum. Since 2007, he has written over 6,000 articles on robotics and technology. He has a degree in Martian geology and is excellent at playing bagpipes.