This fluffy little robot companion will do anything it wants to do. No remote, no app. flatcat reacts to touch instantaneously, playfully explores its environment, and develops individual preferences.
Welcome @ku3i
What a coincidence, someone in the news feed shared an article about your robot:
Can you provide more details about the conception and steps?
Any plan on creating a larger scale version :P?
Curious about the Intrinsic Motivation if you any additional info.
Thanks,
Hi @Mariane,
thanks for the warm welcome.
The idea behind flatcat is to have a fairly simplified robot for the sofa habitat to caress one’s soul. We are currently implementing learning algorithms on flatcat and want the robot to try new things continuously but in harmony with the users who may want to play around with it or may simply want to observe or cuddle it. We also have therapeutic applications in mind. But it’s still at the very beginning.
Intrinsic motivation arises from prediction errors. Over time, the robot builds itself a simple model of itself and the environment (which it experiences via the sensors). Based on the model learned so far, it makes a prediction of the next sensor values. This is not yet perfect, that is, he makes mistakes. But it always gives itself a small reward when it gets the prediction error down a bit. The intrinsic reward. This way it rewards itself for learning and also tries out new things.
Some people love it, some find it kind of creepy. Building it even larger would make the internet freak out, I guess. So maybe we should consider this 
Hi @ku3i
What’s the current size (dimensions) of the FlatCat?
Making it bigger, sounds like a great idea to me 
