Cloud Robotics at Google I/O 2011

On the second day of Google I/O 2011, Ryan Hickman, Ken Conley, Brian Gerkey, and I gave a tech talk about cloud robotics. You can watch the talk now on YouTube.

We'll also be at Maker Faire next week! There will be more robots and plenty more open source goodness.

During the talk, we announced the release of rosjava, the first pure Java implementation of ROS. One exciting aspect of rosjava is Android compatibility. You can now integrate Android devices with your ROS-enabled robots and write apps for them. In addition, with the newly announced Open Accessory API and ADK, you can start controlling actuators or reading external sensors directly from Android devices. Android devices offer tons of sensor and user interface possibilities to robots. Beyond that, they also offer robots a link to the cloud.

Cloud robotics is about making robots universally accessible and useful. Robotics is full of hard problems that make intelligent interaction problematic. Robotics research has gone a long way toward solving many of them. However, even hard problems with solutions become stumbling blocks for new developers who often find themselves reinventing the wheel.

Robot friendly APIs for online services can make state-of-the-art solutions to hard problems universally accessible and allow developers to focus their efforts on making robots useful. Providing these services in the cloud is not only about having access to scalable computing resources. It's also about accessibility. Accessible solutions to hard problems allow students, hobbyists, researches, and professionals alike to combine existing state-of-the-art techniques in new and clever ways to solve even harder or previously unimagined problems.

For example, mapping and navigation are key to mobile robots. Many mobile robots use a technique known as simultaneous localization and mapping, or SLAM, to both learn about their environment and to successfully navigate it. One of the first publications about this technique was written by Smith and Cheeseman in 1986. That was 25 years ago.

The first mass produced, mobile home robot to use SLAM came to market just last year. Although there are numerous specific contributing factors to this delay, I argue that centrally, this is an accessibility problem.

How do we fix this problem? We do it by building a community around open source software and hardware for robotics. Willow Garage has done a great job of this already with ROS. And now, with rosjava running on Android, it's even easier to build awesome robots and robot apps.

No comments:

Post a Comment