This Robot Can Grasp Commands

A team from Cornell University led by an Indian scientist is teaching a robot to understand instructions in its 'language' from various speakers, account for missing information and adapt to the environment.

Ashutosh Saxena, assistant professor of computer science at New York's Cornell University, has developed a software that translates human sentences — such as "Fill a pan with water, put it on the stove, and heat the water. When it's boiling, add the noodles" — into robot 'language'. The robot will also be smart enough to fill in the missing steps.
Saxena's robot is equipped with a 3D camera which scans its environment and identifies the objects in it using the software. The robot has also been trained to associate objects with their capabilities. For example, it knows that a pan can be poured into or poured from; stoves can have other objects set on them and can heat things.
Saxena's team used techniques called machine learning to train the robot's computer 'brain' to associate entire commands with flexibly defined actions.
The computer is fed animated video simulations of the action — created by the scientists in a process that's similar to playing a video game — accompanied by recorded voice commands from several speakers.
The computer stores the combination of many similar commands as a flexible pattern that can match many variations, so when it hears a command it calculates the probability of a match with what it has heard before. If the probability is high enough it associates the two. A similar video simulation supplies the robot with a plan for the action.