A Further Step Toward System Autonomy and Intelligence

Michel Cotsaftis

LACSC/ECE, Paris, France.


More global decision making delegation should be given today man made systems with large number of heterogeneous components and high complexity. A new paradigm is proposed for mix change of power and information fluxes at task control level from identified unit actions for goal accomplishment, with associated trajectory manifold resulting from system self organization. It rests upon system invariants characterizing complex systems dynamics, and uses qualitative methods independent of system state space dimension privileging robustness to compensate for imprecise system parameters and functional variations. A well defined control law is constructed in explicit terms from mathematical (even approximate) representation of system dynamics, by application of fixed point theorem guaranteeing both asymptotic stability and robustness. A “determinism” criterion for autonomous system is considered with evaluation of resulting constraints on system parameters. In this way proposed merging of Mechatronics with Information Technologies provides a first step toward system autonomous intelligence.

Keywords: Complex system, Task control, Fixed point theorem.

     Back to TOC