Hybrid multi-agent framework for real-time knowledge-based control

The approach proposed is intended to take advantage of simultaneous use of agents of different types that, whilst being essentially independent and running mainly concurrently, produce outputs complementing one another in generating the final control decisions. A simple structure of modules cooperation, oriented mainly on a one-directional flow of module activations, enables a high-speed real-time response; a distributed meta-control provides a dynamic adjustment of each module's impact and simplifies possible modifications of the system.<
>