From media streaming to document editing to cell phone backups, everyday computing has never been more reliant on the Cloud – it’s more efficient than ever before, and the robotic world is beginning to feel its benefits. “Cloud Robotics”, a term coined in 2010, describes the concept of network-connected robots taking advantage of parallel computation and data sharing made possible via the Internet.
Big Data – An Infinity of Resources
One can easily see why The Computer Age is synonymous with The Information Age: from simple YouTube tutorials to extensive academic research, every single piece of information ever known to man is available right at our fingertips, thanks to the Internet. Big Data is the term used to refer to the infinitely large library of constantly updated images, maps, and other data on the Web – it’s a tremendous game-changer for robots, which are no longer required to harbor bulky memory space in order to store information locally.
For example, did you know that Google Maps consists of over 21 million gigabytes of data in satellite, earth, and street-level imagery alone? That’s about the storage space of 70,000 average PCs combined. If all of this information wasn’t available on the cloud, the possibility of self-driving would not exist – imagine if your car had to make room for 70,000 hard drives as well as you and your family. Also, since delocalized information on the Internet can be updated in real-time, entire networks of cars can be notified of road closures and other important changes without each car having to undergo an individual system update.
In addition to accessing data, robots can also process the data directly inside the computational environment within the cloud, eliminating the need for localized processing elements. This also allows for interconnection with other computational environments, enabling robots to share data in real-time. Plus, on top of all these benefits, Cloud-based computational environments are highly customizable and secure.
Collaboration and Collective Learning
Almost all cloud based robots work in teams. Robots are far more efficient and powerful when working together – multiple automatic lawn mowers used together to mow a large area, for example. Robots need not necessarily be identical in type to collaborate; in search-and rescue operations, aerial drones and ground robots often work together to provide views and images from different perspectives.
Furthermore, multiple robots can also learn from one another in a process called collective learning. Consider a housekeeping robot designed for picking up various household items off the floor and putting them back where they belong. The most difficult part of this robot’s job would be object recognition and differentiation – unlike humans, who can simply look down and immediately recognize the socks someone forgot to pick up, robots see a two-dimensional image of pixels and colors arranged in a certain way. To figure out what this image represents, the robot utilizes a vast online database of objects, materials, and textures – a database provided by other housekeeping robots employed in different homes. In turn, when this robot learns about a new object not found in the database, it adds its observations to the database so that other robots are able to recognize that object if they ever need to. By aggregating data from all connected robots and applying machine learning technology, robots are able to benefit from crowd-sourced information and increase their functionality.
Big Data, robot collaboration and collective learning are just a few of the numerous advantages offered by cloud robotics. We are in the first inning of the cloud-robotic revolution, where big data, and machine learning are harnessed to rapidly improve functionality and performance of connected robots.