[jeremiah]

LifeAI is an open source C++ library for Artificial Intelligence decision-making. It can be integrated in video games, robotics, or any application that requires artificial intelligence.

Download LifeAI




What is the benefit of LifeAI?

LifeAI helps facilitate entity reactions to the changing dynamics of a complex, virtual or real environment.  The result is a greater sense of autonomy.

What does it include?

LifeAI includes functions for managing variables, and computes action preferences for entities based on the provided variables.  A file input/output system with specialized syntax is also included.  It does NOT include a system for pathfinding, image/audio analysis, or learning (though learning is a future goal).

How does it work?

Entities and actions are created in C++ and text files with LifeAI's syntax.  Entities have characteristics that describe them called "stats".  They consider their value of the stats and how much they are modified to assess the "pros" and "cons" of an action.  They also consider their value of other entities involved in the action and how it affects them.  They can then determine an overall value for performing the action.  As stats in the simulation change, LifeAI updates the best action options for the entities. The actions can then be implemented as needed in a simulation.

Why LifeAI instead of other AI solutions?

LifeAI was designed to be as simple, intuitive, and flexible as possible. It includes optimizations to perform well at real-time speeds.  It can be integrated in almost any project that requires artificial intelligence.

What AI theory is it based on?

It most closely resembles the Utility Theory.

[ivanv]

hi jeremiah

i've surfed over the site. i'm wandering how algorithm for best action decision works?

[jeremiah]

Sure.   First, you may want to refer to this if any of the terminology is foreign:
http://www.lifeai.com/guide_files/general.html#terms
Here is a description of entity logic when actions are considered in laiUpdate():

1. Variable requirements considered, action only occurs if met

2. Entity requirements considered, entities only considered for roles if met

3. Modification amounts calculated
- starts with raw mod amount
- adjusted by mod influencers.
(Influencers are calculated by first determining how far the stat amount is from the universal stat norm.  Once that is obtained, they are divided by the stat norm.  For inverse influencers, the relationship is flipped.  The norm is divided by the stat amount.  Influencers are also scaled by a weight.)
- restricted by stat min/max
- restricted by mod min/max

4. Modification values calculated
- calculate percentage of stat modified
- multiply percentage modified by entity value of stat
- adjusted by mod value influencers
- restricted by mod value min/max

5. Role values calculated
- all of the mod values added together for the role
- entity in a particular role looks at the role values of all roles, scales them according to their value of the entity occupying the role, adds the resulting scaled and polarized role values together to form their final role val.
- restricted by role value min/max

[jeremiah]

Version 1.1 is now available. This release focuses on usability enhancements.  Highlights of the release include:

- Actions with negative values can now be selected as the "best action". Enables entities to choose the "lesser of two evils," if an action with a positive value is not an option.
- Added support for new "string" type. Convenient for storing words or phrases in LifeAI files.
- Added feature for creating multiple instances of variables, stats, strings, entities, and actions in LifeAI files at the time of their creation.
- Added laiSetLink() and laiSync() functions as an alternative way to sync data to and from LifeAI. Allows linking the memory address of a data type such as a floating point number to a variable, stat, or string in LifeAI. When laiSync() is called, syncs changes in the data to LifeAI.