Playing mario vs donkey kong using Universal Artificial Intelligence
Note: To make this website free to the public
please click on an ad to support my sponsors or you
can make a tax-deductable donation using Paypal
(click on the donation icon on the left).
This video shows a robot
solving a puzzle game. There are no sound in this
video because I wanted the viewers to focus on the
robot's thoughts while solving a problem.
When the robot does a
task, such as play a videogame, there will be
recursive tasks the robot has to do. In this example,
the robot is trying to find a way to reach mario
junior. There are several problems the robot is
confronted with to reach mario jr. He uses logic and
common sense in order to solve these problem. Each
problem the robot wants to solve is considered a
sub-task of the overall task. Within each sub-task
are even more tasks.
To make things even
more complex, the robot has no knowledge about the
objective of the level or the rules of the level. He
has to use common sense knowledge in order to
discover the strategies to pass the level. In the
video, I show that the robot is using some ideas that
leads to dead ends. After numerous trial and error,
the robot finds the linear steps to navigate to mario
jr.
If you analyze this
video further, the robot is using a much more
sophisticated way of solving problems than recursive
tasks. Tasks are interconnected with each other,
previous tasks done are revisited, some deep
recursive tasks are forgotten, some tasks are bypass
because of better methods found, etc. The robot's
conscious creates an optimal computer program to
manage tasks. This computer program manipulates task
in the task container using different data structures
(not just recursive tasks). The more the robot plays
a level the faster the level can be solved.
