I was in Berkeley a couple of months back, helping TechCrunch get its proverbial ducks in a row before our first big climate event (coming in a few weeks, folks). Professor Ken Goldberg, who’s helped play a big role in a number of our robotics events, kindly gave me a tour of BAIR, the Berkeley Artificial Intelligence Research department. It’s a newly renovated space on the UCB campus, where students and faculty are working on some cutting-edge robotics research.

I got previews of a number of projects I hope to share with you in the newsletter soon, but one that really caught my eye was FogROS, which was just announced as part of the latest ROS (robot operating system) rollout. Beyond a punny name that is simultaneously a reference to the cloud element (fog/cloud -- not to mention the fact that the new department has killer views of San Francisco and frequent visitor, Karl) and problematic French cuisine, there’s some really compelling potential here.

I’ve been thinking about the potential impact of cloud-based processing quite a bit the last several years, independent of my writing about robots. Specifically, a number of companies (Microsoft, Amazon, Google) have been betting big on cloud gaming. What do you do when you’ve seemingly pushed a piece of hardware to its limit? If you’ve got low enough latency, you can harness remote servers to do the heavy lifting. It’s something that’s been tried for at least a decade, to varying effect.

Image Credits: ROS

Latency is, of course, a major factor in gaming, where being off by a millisecond can dramatically impact the experience. I’m not fully convinced that experience is where it ought to be quite yet, but it does seem the tech has graduated to a point where off-board processing makes practical sense for robotics. You can currently play a console game on a smartphone with one of those services, so surely we can produce smaller, lighter-weight and lower-cost robots that rely on a remote server to complete resource-intensive tasks like SLAM processing.