BECCA (Brain Emulating Cognition and Control Architecture) is a computer program created by Sandia Labs researcher, Brandon Rohrer. Rohrer believes that creating an AI with a canine level of intelligence could happen within the next ten years. BECCA is a brain-emulating cognition and control architecture he hopes will reach this level. The program gives the robot the ability to learn from its experience and to develop very simple problem solving strategies. Rohrer describes his approaches to the reinforcement learning problem, "deep learning", and how a human level AI might be created within the next twenty years.
According to Rohrer:
BECCA is a computer program, a robot brain that can learn from its experiences and create abstract concepts, in the roughly same way that children do. I hope that one day it will help robots do everything I can do, including talk with people, climb mountains, clean the kitchen, and build other robots. But I'll be happy if it helps a robot to be as smart as a dog. I have chosen to cast the challenge of natural world interaction as a reinforcement learning (RL) problem: an agent takes actions and receives sensory information at each time step. Its goal is to maximize its reward. The problem formulation assumes nothing else about the nature or structure of the environment. In order to address this general RL problem, I develop biologically-motivated algorithms and incorporate them into a Brain-Emulating Cognition and Control Architecture (BECCA).
The video below shows BECCA operating on robots with different configurations. (We're guessing his kids helped with the production).
Below: brief description of BECCA given at the Artificial General Intelligence Conference, held at Google in Mountain View, California on Aug 4, 2011. The discussion panel that followed the session contained some additional information on the topic.
If you are running MATLAB, you can download a free copy of BECCA from the link below.
In the future, BECCA will be recompiled to Python, so that it can be integrated into the ROS language.