University of Queensland (UQ) postdoctoral research fellow Dr Ruth Schulz and her colleagues have created a pair of robots that have their own language.
Credit: University of Queensland
The ‘Lingodroids' are a pair of mobile robots that communicate by developing their own words for places, and relationships between places based on distance and direction. The Lingodroids are a pair of mobile robots that evolve a language for places and relationships between places (based on distance and direction). Each robot in these studies has its own understanding of the layout of the world, based on its unique experiences and exploration of the environment.
Credit: University of Queensland
Despite having different internal representations of the world, the robots are able to develop a common lexicon for places, and then use simple sentences to explain and understand relationships between places – even places that they could not physically experience, such as areas behind closed doors. By learning the language, the robots are able to develop representations for places that are inaccessible to them, and later, when the doors are opened, use those representations to perform goal-directed behavior.
The language sounds like a sequence of phone tones, which are easy for the robots to produce and hear in a noisy office environment, before being translated into syllables to make it easy for humans to recognise them.
Dr Schulz said that the robots start by playing where-are-we games.
“If they encounter an area that has not yet been named, one will invent a word, such as “kuzo”, choosing a random combination of syllables, which it is then able to communicate to other robots it meets, thus defining the name of the place,” she said.
“These words are known as “toponyms” (“topo” meaning place and “nym” meaning name).
“The robots then start to play how-far and which-direction games, which enable them to develop relationship words (like English prepositions).”
The resulting language consists of location, distance and direction words, enabling the robots to refer to new places based on their relationship to known locations.
“These languages are very powerful – they are known as “generative” languages because they enable the robots to refer to places they haven't been to or even places that they imagine beyond the edges of their explored world,” Dr Schulz said.
An essential aspect of these games is that the robots develop robust ideas of where a word should be used.
Their understanding of the new language was tested using games in which two robots attempted to meet at a particular toponym, or place name.
If one robot told the other “jaya”, they would independently navigate to where they thought “jaya” was.
When both robots arrived at the same location, the concept “jaya” was consistent between the robots.
After having played hundreds of games to develop their language, the robots agreed upon concepts for toponyms within 0.65 metres, directions within 10 degrees and distances within 0.375 metres.
The robots consist of a mobile platform which is fitted with a camera, laser range finder, sonar for mapping and obstacle avoidance, and a microphone and speakers for audible communication with each other.
Dr Schulz and her colleagues presented their research at the International Conference on Robotics and Automation in Shanghai on Tuesday, May 10, and have since received international coverage.
“We believe that the natural way to communicate with robots will be through human language,” Dr Schulz said.
“The Lingodroids have developed their own proto-language as part of realising this ambition,” she said.
“In the future, our aim is to extend the types of concepts able to be formed by the robots, as well as expand to additional grammatical features of language and to human-robot interaction.”
Source: University of Queensland
Lingodroids
The ‘Lingodroids' are a pair of mobile robots that communicate by developing their own words for places, and relationships between places based on distance and direction. The Lingodroids are a pair of mobile robots that evolve a language for places and relationships between places (based on distance and direction). Each robot in these studies has its own understanding of the layout of the world, based on its unique experiences and exploration of the environment.
Despite having different internal representations of the world, the robots are able to develop a common lexicon for places, and then use simple sentences to explain and understand relationships between places – even places that they could not physically experience, such as areas behind closed doors. By learning the language, the robots are able to develop representations for places that are inaccessible to them, and later, when the doors are opened, use those representations to perform goal-directed behavior.
The language sounds like a sequence of phone tones, which are easy for the robots to produce and hear in a noisy office environment, before being translated into syllables to make it easy for humans to recognise them.
Dr Schulz said that the robots start by playing where-are-we games.
“If they encounter an area that has not yet been named, one will invent a word, such as “kuzo”, choosing a random combination of syllables, which it is then able to communicate to other robots it meets, thus defining the name of the place,” she said.
“These words are known as “toponyms” (“topo” meaning place and “nym” meaning name).
“The robots then start to play how-far and which-direction games, which enable them to develop relationship words (like English prepositions).”
The resulting language consists of location, distance and direction words, enabling the robots to refer to new places based on their relationship to known locations.
“These languages are very powerful – they are known as “generative” languages because they enable the robots to refer to places they haven't been to or even places that they imagine beyond the edges of their explored world,” Dr Schulz said.
An essential aspect of these games is that the robots develop robust ideas of where a word should be used.
Their understanding of the new language was tested using games in which two robots attempted to meet at a particular toponym, or place name.
If one robot told the other “jaya”, they would independently navigate to where they thought “jaya” was.
When both robots arrived at the same location, the concept “jaya” was consistent between the robots.
After having played hundreds of games to develop their language, the robots agreed upon concepts for toponyms within 0.65 metres, directions within 10 degrees and distances within 0.375 metres.
The robots consist of a mobile platform which is fitted with a camera, laser range finder, sonar for mapping and obstacle avoidance, and a microphone and speakers for audible communication with each other.
Dr Schulz and her colleagues presented their research at the International Conference on Robotics and Automation in Shanghai on Tuesday, May 10, and have since received international coverage.
“We believe that the natural way to communicate with robots will be through human language,” Dr Schulz said.
“The Lingodroids have developed their own proto-language as part of realising this ambition,” she said.
“In the future, our aim is to extend the types of concepts able to be formed by the robots, as well as expand to additional grammatical features of language and to human-robot interaction.”
Source: University of Queensland
The Lingodroids can be seen playing a location language game at www.youtube.com/watch?v=vhUOUaSNP3w
More information about the research can be found at the Lingodroids web site,http://itee.uq.edu.au/~ruth/Lingodroids.htm
More information about the research can be found at the Lingodroids web site,http://itee.uq.edu.au/~ruth/Lingodroids.htm
Schulz, R., Glover, A., Milford, M., Wyeth, G., & Wiles, J. (2011) Lingodroids: Studies in Spatial Cognition and Language, ICRA 2011, The International Conference on Robotics and Automation, Shanghai, China, May 2011
0 comments:
Post a Comment