Hyderabad: We know robots are the future. But humanoids take a lot of time learning to do their tasks. Hyderabad-raised researcher Dr Rajesh P.N. Rao and his team has found a novel way of teaching robots using crowd sourcing technology. They say robots can just go online and find hints from hundreds of people to do their tasks by themselves without having to teach them with painstaking long commands.
In a first, University of Washington associate professor Dr Rajesh P.N. Rao and his team had demonstrated the first brain to brain interaction through the Internet in which one person would imagine an action and another ‘online’ would replicate the imagined action. Working on robots, Dr Rao, who studied in Hyderabad till the age of 18, says humanoids could one day proliferate in our environment.
But robots are now designed to do a particular task or a set of them. They have to be programmed to teach them how to go about a job. Robots learn by imitating humans, but it could take many lessons for it to hold a dish perfectly or place the book at the right place.
But Dr Rajesh Rao’s team has proposed using crowd sourcing techniques where robots can, all by themselves, go online, and ask for advice “Robots in the not-so-distant future will be in our homes, offices, and other human environments. We will want them to learn new skills on-the-job so that they can get better at helping us” Dr Rajesh Rao said.
"Our research demonstrates that robots can go online, and hire human helpers whenever they are stuck and get hundreds of hints and suggestions from humans all around the world," he further said.
Crowd sourcing is the process of soliciting information from a large online community. "This is a way to go beyond just one-on-one interaction between a human and a robot by also learning from other humans around the world," Dr Rajesh said.
However, there could be problems if some spammers in the online community feed wrong information to the robot. "There were some spammers who produced repeated data from multiple accounts, and even others who produced data without following instructions. We are planning to adopt more complex quality control techniques to improve the quality of crowd sourced data in future," researchers said.
