DirectIA®: Autonomous Behavior Kernel

Thanks to this proprietary technology, MASA-SCI develops virtual or constructive simulations showing unprecedented realism.

DirectIA®" stands for " Direct Intelligent Adaptation". It is the world's first generic artificial life & behavior SDK. DirectIA® was built to create agents that adapt to unpredictable situations (situations which were neither predicted nor programmed), look alive and seem to have intelligent behaviors.

The patented technology behind DirectIA® behavioral engine rests on a fusion between biology, computer science and mathematics. It is the most advanced technology on the market, supported by a world-class development platform.

DirectIA capabilities

The growing need to train and stimulate decision makers has put a specific emphasis on simulation as an efficient alternative to costly full scale exercises. However simulation solutions still suffer from critical drawbacks:

  • They demand a very high level of manning. Those system controllers are mainly push button operators and waste their time.
  • Their realism is sacrificed and simulated situations are usually oversimplified.

In this respect, DirectIA® represents a major breakthrough in the simulation domain. It has been designed and developed to allow for the representation of realistic and complex situations involving simulated entities behaving autonomously or according to a doctrine, understanding operational orders and sending back operational reports.

Close to operational work

DirectIA® simulated entities all share the same characteristic: they speak ‘operational’. All interactions are made using operational language and procedures, as close as possible to the real way of commanding. As a result, the number of system controllers required by legacy simulation systems implementation is dramatically decreased since they no longer need to translate orders issued by the decision makers into low level actions understandable by the simulation core. Decision makers can also directly use DirectIA® powered solution on a single laptop, bringing in a very flexible manner simulation capability within a command post.

DirectIA ® does not alter the way those people work, DirectIA® simply replaces a real life situation by a virtual situation.

Example:

DirectIA® models implement simulated Companies & platoons capable of literally interpreting operational mission instructions, in the same way an officer would implement doctrine and former training. Mission term, boundaries, phase lines, and objectives are directly taken into account to set up a doctrine-compliant course of action.

  • Orders issued by Battalion commanders during a Brigade command post training session are directly taken into account by simulated Company commanders thus dramatically limiting the number of captain level push button operators.
  • Battalion commanders can also test and optimize the courses of actions they are elaborating by themselves: they simply need to draw their own orders along with the opposing forces supposed course of action and let the models act.

Doctrine and autonomous behaviors modeling

DirectIA® makes simulated entities capable of reacting to various environment stimuli, ranging from users interactions to unexpected events. Doctrine, automous behaviors and human factors can be incorporated in classical models.

Consequently, entities, expected to follow a doctrine, exhibit consistent behaviors while other entities follow their own motivations.

Example:

  • DirectIA® Companies & platoons are capable of carrying out autonomously more than 600 hundreds missions.
  • DirectIA® smart opponent included in the Real Time Strategy videogame Conflict Zone (Ubisoft) exhibits two key features: it automatically tries to reach its own goals while adapting its behaviors to the decisions made by the gamer, whose tricks and strategies will work only once…

Efficient Update process without reengineering

The doctrine and threats to security will obviously continue to evolve implying that models are bound to be updated more and more often to remain accurate and relevant. DirectIA unique design turns model update into a simple and efficient process. Advanced users basically implement the new course of action, test it and validate it. DirectIA® ensures that all the already existing courses of actions and validations and verifications remain valid.

This has to be compared with the classical update process which brings software engineers to integrate the new course of action and then test and validate all the existing courses of actions to ensure the stability of the whole model.

Complexity management

First applications of simulation represented simple environments and put a specific emphasis on physical models such as attrition and detection. This brought the M&S frameworks and technologies to ask for the description of every possible case and corresponding behavior. Theoretically, such approaches obviously work but practically, with the growing complexity of modern crisis, they face combinatorial explosion.

On the other hand, DirectIA approach simply implies the description of the response to each environment stimulus, its algorithms being precisely designed to manage the competition between those responses. Modeling workload is thus basically linear with regards to the environment complexity.

Focus on decision modeling

DirectIA entity

A DirectIA® entity is a classical simulated entity enhanced with reasoning capabilities. While the standard simulations basically address the ‘how”, DirectIA® focuses on the ‘why’. A DirectIA® entity is thus described by its physical assets and its decisional principles.

The behaviors database is available for modifications by both technology users and DirectIA® based simulation end-users who can, depending on their level of training:

  • Customize generic behaviors libraries to account for specificities or create entities behaving differently
  • Refine existing behaviors to enhance realism
  • Add new behaviors to remain compliant with an evolving doctrine or behavioral corpus
  • Dedicated tools support the behaviors database edition process.
  • Reasoning capabilities available include:
  • Elaboration and update of adapted responses to environment stimulations
  • Ability to lead and command
  • Human factors influences on responses

World representation

This module gives DirectIA® entities the capability to build their own representation of the world they live in. This representation is the combination of what entities see and what entities know. They mix the results of their own visual detections, information gathered through interactions with other entities and their own knowledge.

This representation serves as a basis for entities responses to environment stimulations.

Hierarchy

This module adds functionalities to the DirectIA® engine enabling entities to communicate among them by sending information. Depending on its type, information has several impacts:

Update of entity representation of the world.

Potential modifications of entities behaviors belonging to a group. Representation of group leaders who influence other individuals.

Drastic modification of entity behaviors. Representation of a chain of command and entities capability of issuing orders to other entities.

Human factors

The motivation module provides a special kind of goal-generation mechanism called a motivational system and inspired by ethological studies. It is based on a simple biological model: an entity has internal needs (eating, sleeping, etc.) and is submitted to external constraints (resources available in the environment). When an agent's needs are not satisfied, they create internal disequilibrium. Thus, the permanent goal of the entity is to restore its internal equilibrium.

Mathematical models of human factors can be integrated within this framework whose architecture is fully open. They will directly influence motivations of an entity and in turn modify the way it implements course of actions.

Example:

è DirectIA® companies & platoons ways of implementing an attack mission may depend on their morale. Even a very simple model implying that a low morale reduces the courage can help higher level command posts apprehend impacts of psychological states of subordinated units on maneuvers. Such experimentations have been carried out at the French Simulation Office and deeper studies are conducted to introduce such phenomena during official Brigade and Division command posts training sessions.