HUMANS (HUman Models based Artificial eNvironments Software platform) brings together all the models and decision engines developed by the HUMANS team. Its purpose is to support the modelling and scenarisation of human activity in complex virtual environments. The architecture of this software suite was designed collaboratively by the PhD students and research engineers of the HUMANS team.

Figure: HUMANS architecture

Models and modules HUMANS is made up of modules, each with a well-defined function. These modules are built around models that provide them with shared representations. For clarity, some of them have been renamed, and some modules have been functionally split. The five main modules of HUMANS are:

• HUMANS»WORLD MANAGER: this module manages the state of the world. It determines whether actions succeed and what their effects are. It corresponds to the former COLOMBO;
• HUMANS»REPLICANTS: this module handles the decision-making models of autonomous virtual characters (Margot Lhommet);
• HUMANS»MONITOR: this module is responsible for learner monitoring and activity traces. It corresponds to the trace-related part of HERA (Kahina Amokrane);
• HUMANS»SELDON: this module is responsible for scenarisation. It corresponds to the scenarisation part of HERA, extended with the control of autonomous virtual characters. It comprises two sub-modules:
  • HUMANS»SELDON»TAILOR: this module generates the scenario according to the learning objectives and the learner's profile. It produces a template scenario structured by milestones (Kevin Carpentier);
  • HUMANS»SELDON»DIRECTOR: this module dynamically steers the scenario according to the learner's actions and the milestones set by TAILOR (Camille Barot);
• HUMANS»COMMUNICATOR: this module is the interface between the HUMANS modules and external components (e.g. a 3D engine).

The HUMANS models are the formalisms used for knowledge representation. They form the common foundation of all modules and were therefore designed by the whole team. They are grouped under the name HUMANS»HUMANS-DL (Description Language). The three main models are:

• HUMANS»DOMAIN-DL (domain model): this model represents the entities of the world and how they behave;
• HUMANS»ACTIVITY-DL (activity model): this model represents goal-directed activity (formerly HAWAI-DL);
• HUMANS»CAUSALITY-DL (causality model): this model represents the causality between events, their consequences and the associated probabilities. It extends the risk model to events other than risks.

A graphical editor, HUMANS»EDITOR, allows each of these models to be edited. Each model has its own dedicated view, which is used to enrich and specify the entities of the world as well as the associated activities and causal relations (formerly Visual HAWAI).