Imagine living life without 24/7/365 access to Google. Better yet, imagine not being able to instantly do any of the following things:
- Look up the weather forecast for any city in the world.
- Read dictionary definitions for any word.
- Find nearby restaurants and grocery stores.
- Read local and national news stories.
- Make arithmetic calculations.
- Find out when a movie is dropping.
- Convert metric to standard and vice versa.
Now you know what it feels to be a simpleton robot. Yep. Robots might possess super-fast processors and large memory chips, but they lack the brilliance of Google — or at least they used to lack it.
Enter Robo Brain, or what Gizmodo contributor Pranav Dixit describes as “a large-scale computational system that learns from publicly available Internet resources.” This system is currently “downloading and processing” everything it can get its hands on: images, YouTube videos, how-to documents, appliance manuals, etc. All this data will then be stored on the Robo Brain database, where robots from across the world will be able to access it when required.
Confused? Ashutosh Saxena, assistant professor of computer science at Cornell University, explained it in a EurekAlert! news release: “Our laptops and cell phones have access to all the information we want. If a robot encounters a situation it hasn’t seen before it can query Robo Brain in the cloud.”
Cornell Chronicle contributor Bill Steele provides a bit more insight:
“If a robot sees a coffee mug, it can learn from Robo Brain not only that it’s a coffee mug, but also that liquids can be poured into or out of it, that it can be grasped by the handle, and that it must be carried upright when it is full, as opposed to when it is being carried from the dishwasher to the cupboard.”
Get it now? Robo Brain will be like a cloud-based Google that robots (be it it robot vacuum cleaners, robot lifting aides, robot bomb disposers, etc.) can use to look up information. Pretty cool, right?
So how does it work? Robo Brain works via what scientists refer to as “structured deep learning.” Information is stored in a way similar to the way data is stored in object-oriented languages. For instance, an office chair would be an object of the chairs class, and the chairs class would be an object of the furniture class. Keep in mind that this is a very simplistic example that ignores the incredible complexity of this sort of information storage.
According to the Cornell Chronicle, “a robot’s computer brain stores what it has learned in a form mathematicians call a Markov model, which can be represented graphically as a set of points connected by lines (formally called nodes and edges).” Each node is an object, action or image, and each node “is assigned a probability” based on “how much you can vary it and still be correct”:
“In searching for knowledge, a robot’s brain makes its own chain and looks for one in the knowledge base that matches within those probability limits.”
Confusing, right? Regardless, to learn more about this incredible project, be sure to check out the official Robo Brain website!