Overview of PELEA Architecture

The architecture for PELEA includes components that allow the applications to dynamically integrate planning, execution, monitoring, replanning and learning techniques. In general, there are two main types of reasoning: high-level (mostly deliberative) and low-level (mostly reactive). In our architecture, these are simply two planning levels. This offers two main advantages:

• both levels can be easily adapted to the requirements of the agent;
• and the differentiation allows replanning at either level, which grants a greater degree of flexibility when recovering from failed executions.

It would be possible to add additional levels to allow developers for a more hierarchical decision process. However, we consider that the sole distinction between high and low level is enough to tackle most problems, as has been shown in many robotics applications.

PELEA is composed of eight modules that exchange a set of Knowledge Items (KI) during the reasoning and execution steps.

Next: Where to use PELEA ?