Robo-Culture: This is what my 4 year project is about. Finding ways to give robots an artificial culture so that when we, as humans, interact with them it will feel as though the robots come from a culture. By coming from a culture, the robot will have a character of its own, which would produce interesting patterns of behaviour.
At the root of creating a robo-culture is creating an artificial society. Inside this society, robots can undergo social interactions with themselves. One of the sociological theories that this artificial interaction is based around is the Cooley Looking Glass Self. By using this, we enable the robots to take on opinions, biasing them to different actions.
Summer 2009 at Clarkson University I did Honors summer research and transferred this idea of using an artificial society to socialize social robots into code. This program, when iterated from 0-1000 times, showed that the bundles of AI algorithms implemented allowed a child robot to experience periods of likeness and dislikeness towards its parents, similar to what happens in real life.
If you’re interested, I encourage you to download the paper and read it.
Socializing a Social Robot with an Artificial Society
Social behaviours have been developing within our world for centuries. The balance of many variables simultaneously plus a history of social interaction shapes the way we interact with others. In order to allow robots to gain social behaviour in order to interact with others, a process whereby the robot undergoes the experience of socialization is necessary. Key factors of the socialization experience include the culture of the society, the updating and changing of beliefs, and determining favourable actions. These key factors were implemented in a software environment by creating an algorithm that mimics the Looking-Glass Self theory. Additional probabilistic methods were implemented to further extend the depth of the robo-culture within the artificial society. Results of the created algorithms show consistent patterns within the robot’s behaviours.
Click here to download the paper (pdf)
Click here to go to the Zoomify graphs page
The code has now been subdivided into two parts, Control and the Drawing. Control is where all the algorithms are crunched, and Drawing draws information. The code is still in its introductory state, but rather than hiding it from the world, I have decided to share it! Please note that this will not provide an instant implementation of sociable robots. There are still many stages of the robot’s life and evolution that need to be coded, documented, and brainstormed. With that said, here is the code! It is licensed under the CC BY-NC-SA License.
As of posting this, there are two main things that need to be improved and will be improved in the following days:
Processing in Netbeans (Porting Processing to Netbeans)– Done! (For the most part)
- Documentation (I have two notebooks full of pictures and flowcharts that deserve to be documented)
Still interested in more of the project? Here’s some reading material!
- 3rd Annual Undergraduate Research Symposium Dessert Reception at Clarkson University, Fall 2009 [Poster Presentation]
- Symposium on Undergraduate Research (SURE) at Clarkson University, Summer 2009 [Poster Presentation]
- PechaKucha Night #12 at Montreal, QC
I focused my efforts on the Social Robotics project through the Clarkson Open Source Institute (COSI) project class, MP151, for 2 credits.
The main goal in Fall 2009 was to enable the program to run without crashing (it previously crashed around 1500 iterations) and draw the conformity levels. Once these were accomplished, a very peculiar observation was made whereby the pattern that existed from 0-1000 iterations did not stretch to 0-100,000 iterations!
Various documentation during the project:
There is literally a zillion characters of documentation from the summer since we had to send in detailed weekly reports. This will be posted at a later day, keep checking back!