May 24, 2011

Researchers Developing Lingodroid Robot Language

Researchers are developing a language specifically for robots to help them navigate and improve themselves.

The Lingodroid research project allows robots to generate random sounds for the places they visit in both simulations and a real office.

These sounds are shared and the robots play games to help establish which sound represents which location.

The language is so sophisticated that it can be used to help robots find places other robots direct them to.

Project leader Dr. Ruth Shulz from the University of Queensland said the robots are generating their own words because human language is loaded with information that robots do not understand.

"Robot-robot languages take the human out of the loop," she said in a statement. "This is important because the robots demonstrate that they understand the meaning of the words they invent independent of humans."

One set of the trials sees wheeled robots fitted with a camera, laser-range finder, and sonar used to map their world.  The robots have a microphone and speakers onboard so they can communicate with each other.

When a traveling robot reaches a place that does not have a name, it generates a random combination of syllables that represent that place.

When a robot meets another robot it informs it of the places it has been.  Some of the names generated by the robots are "kazoo", "jar" and "fexo."

Schulz said each location was broadly tied to the sensory horizon of the sonar and laser-range finder they have on board. 

She said each chunk of territory was a few feet in diameter, which enables names to be used as a round distance measure.

Shulz said the power of the language was starting to become apparent.

"They enable the robots to refer to places they haven't been or even places that they imagine beyond the edges of their explored world," she said in a press release.

The team wrote about their research in the paper "Lingodroid: Studies in Spatial Cognition and Language."  They presented the research last week at the IEEE International Conference on Robotics and Automation in Shanghai, China.


On the Net: