AI, soft- and hardware development

Feedforward closed loop learning

Feedforward closed loop learning (FCL) is a learning algorithm which adds flexibility to autonomous agents

It's been designed for non-player-characters in gameplay: A designer defines a baseline behaviour and then FCL learns to improve it

Vizdoom demo

At 0:30 FCL is activated to improve the behaviour and is then able to successfully shoot the enemy


FCL is available under the GPL v3 on github


Contact us for a closed source license

Contact us