In what could be a scene straight out of a science fiction blockbuster, scientists at Cornell University in New York are now developing what could be a quantum leap forward in service-oriented robots.
“Robo Brain” is a large-scale computational system that uses publicly available resources to teach itself how humans interact with their environment and the various elements in it. The cloud-based platform is being loaded with about 1 billion images, over 100,000 Youtube videos, and hundreds of millions of DIY, how-to, and appliance manuals. “Our laptops and cellphones have access to all the information we want. If a robot encounters a situation it has not seen before, it can query ‘Robo Brain’ in the cloud,” explained lead researcher Ashutosh Saxena. This ability to access relevant data in the cloud should enable robots equipped with a connection to Robo Brain to be more effective and efficient assistants for their human masters.
For example, a robot that “sees” a coffee mug for the first time can learn from Robo Brain exactly what it is, including pertinent details like the fact it can be used to hold liquids, can be held by its handle, and must be upright when full in order to avoid spills. A robot that “sees” a non-working light bulb will “know” how to remove and replace it with the correct bulb, without the need for a human (or machine) to program the multiple steps needed to accomplish the task.
“Robo Brain will look like a gigantic, branching graph with abilities for multi-dimensional queries,” explained Aditya Jami, visiting researcher at Cornell who designed the large-scale database for the brain.
The list of possible uses for this evolving technology is virtually limitless. Robots can be sent to repair just about anything and be “trusted” to know how to make that repair simply by looking at the item. On the other hand, teams of robots can be sent out with blanket instructions to fix every broken widget they encounter.
Makes one wonder about the viable future of the Maytag Repairman.