TurtleBot 2 Prototype Unveiled at ROSCon

We meet the new successor to Willow Garage’s low cost, open source robot platform

4 min read

Evan Ackerman is IEEE Spectrum’s robotics editor.

TurtleBot 2 Prototype Unveiled at ROSCon

Willow Garage introduced the TurtleBot in April of 2011 as a low-cost personal robot kit running open-source software. A year later, it's been an incredible success with researchers and hobbyists, and it's time to meet the next generation platform. Fans of the original TurtleBot needn't worry: TurtleBot 2 is intended to be just like the TurtleBot you already know (mostly) and love (when it works), except with a brand new full-featured base and a sinister flat-black color scheme.

So, why do we even need a TurtleBot 2? Besides the fact that newer and more capable robots are awesome and fun, there are all kinds of practical reasons why the original TurtleBot needed a successor. The biggest one is that iRobot Create bases (and, consequently, TurtleBots themselves) can’t get exported outside of the U.S. because there’s no electronics certification and also because Create can’t handle 220-volt power supplies (don’t even try it, you’ll completely fry your TurtleBot as soon as you plug the poor thing in). So, in order for TurtleBots to be distributed in Europe and Asia, a new non-Create mobile base was necessary.

Yujin Robot, a South Korean domestic robotics company (they make vacuums, among other things), decided to try and make this happen. While they were at it, they asked Willow Garage, Clearpath Robotics, and I Heart Engineering (designers and manufacturers of TurtleBot) what was especially annoying about the Create platform so that Yujin could create a next-generation base that incorporated some substantial improvements. Now, there’s nothing wrong with Create, per se, it’s just that it’s a 10 year old platform at this point, and 10 years is an incredibly long time in the world of robotics.

Obviously, Yujin wanted to update TurtleBot with a base that could be used and sold anywhere, but they also added all kinds of features that will make the TurtleBot 2 way, way more turtlier than its predecessor. The new base itself is called Kobuki, and here are the major differences between Kobuki and the original Create:

  • Wheel encoders: an odometry system (11.7 ticks/mm) offers much more accurate localization.
  • Integrated gyroscope: Kobuki includes a factory calibrated 1 axis (100 deg/s) heading gyro.
  • Big batteries: in addition to a larger integrated battery, Kobuki has an extra battery pack on top, giving you about 2-3 hours of reliable life (a total of 6600 mAh capacity).
  • Better hardware access: all the connectors you could want (for power and communication) are easy to access through a panel on the back of the robot, and if you want to burrow down farther, you can still get pin access.
  • More power: Kobuki can deliver more power to laptops, sensors, manipulators, and other accessories.
  • More speed: the spec sheet for Kobuki says it tops out at 50 cm/s, but from what I’ve seen, it’s capable of moving significantly fast­­­­­­er than that, while staying quieter than the Create. Also, at the moment it goes way faster backwards for some reason.
  • Everything else: like Create, Kobuki comes with bump sensors (separated into left, right, and center channels), cliff sensors, wheel drop sensors, as well as programmable audio, LEDs, and touch buttons.


The other big thing that Kobuki will offer is, hopefully, the ability to use a self-charging dock. This is big for two reasons: obviously, there’s the convenience of not having to plug your robot in all the time, but more importantly, it makes continuous operation possible. With the dock charging Kobuki, and Kobuki passing power through to the laptop, you’ll be able to leave TurtleBot 2 on 24/7, which enables true remote access for robot sharing or telepresence or unsupervised autonomy.

From the base up, TurtleBot 2 is drop-in compatible with the TurtleBot 1. If you already own a TurtleBot 1, you can buy a new Kobuki base, and switch it out with the Create. The only notable changes that you’ll see buying a complete TurtleBot 2 is a more powerful laptop (since whatever laptop or netbook is currently available will by default be more powerful than the ones that came with earlier TurtleBots), and likely the substitution of an Asus Xtion sensor for the Kinect, which may offer slightly higher performance.

Kobuki will certainly offer hobbyists a lot more features and options in the next gen TurtleBot, but arguably, Yujinn (and Willow) are trying to start moving beyond hobbyists (an inclusive beyond, mind you) with this new release. Yujin sees the beginnings of a big push in Asia towards robotic education, but without a common hardware (or software) platform, it’s very difficult to implement a robotics curriculum. Unfortunately, TurtleBot 1 wasn’t quite good enough for most classroom applications for a variety of reasons, mostly related to the Create base. The hope is that with the TurtleBot 2 eliminating a lot of the frustrations of the platform, you’ll be able to buy one (or a bunch) and run a TurtleBot class without having and specialized robotics knowledge.

The idea of a curriculum that could be taught in a school by a teacher who isn’t a roboticist is something that’s also been on the mind of Clearpath Robotics, and they’re actively working on a bunch of resources to make it easy to just drop a fleet of TurtleBots into a classroom and get them up and running. Clearpath CEO Matt Rendall talks more about that in the following video, which also gives a first look at the TurtleBot 2 directly from ROSCon and includes some absolutely brutal TurtleBot-on-TurtleBot violence:

The TurtleBot 2s we met at ROSCon were just prototypes; there are still some tweaks that will likely be made to the design as Yujin gets feedback from the people at Willow and Clearpath and I Heart Robotics. After that, Yujin will be ramping up to pre-production, with the goal of getting the first TurtleBot 2s out the door in September of this year. As for the price, it should be about the same as it would cost you to buy a TurtleBot 1 (with the Create base) and add a power board and gyro to it, so think something like $1,500ish. Not cheap, we know, but not impossible for determined individuals to afford either, and remember, you're getting a very clever and capable robot backed by a robust open-source operating system and a community to support it (and you) every step of the way.

[ Kobuki ]


The Conversation (0)