A class of BDI agent architectures for autonomous control

A class of beliefs-desires-intentions (BDI) agent architecture is presented for control systems with a high degree of autonomy. The architecture contains agents for modelling, controller optimization, implementation and to monitor performance. The global convergence of performance of the agent system is proven under three mild assumptions. Relevant features of the agent structure are competing modellers and controllers. The benefit is an enhanced ability to learn new plant dynamics of varying complexity and controller adaptation. The new family of control agent architectures is called cautiously optimistic, a name to reflect the most important property of the new architecture: modelling results are applied with caution for control but current models are accepted until measurements do not contradict them with a margin. A cautiously optimistic control agent (COCA) is proven to have converging performance to a nearly optimal performance for stationary dynamics of a real plant under fairly general assumptions.

[1]  S. Veres.  Self-tuning control by model unfalsification (Part I) , 2000 .

[2]  Sandor M. Veres,et al.  Improved efficiency of adaptive robust control by model unfalsification , 1999, Autom..

[3]  S. M. Veres.  Self-tuning control by model unfalsification (Part II) , 2000 .

[4]  S. Veres.  Iterative identification and control redesign via unfalsified sets of models: A basic scheme , 1999 .

[5]  Robert L. Kosut.  Iterative Adaptive Robust Control via Uncertainty Model Unfalsification , 1996 .

[6]  Peter E. Caines,et al.  SPECIFICATION AND SUPERVISORY CONTROL FOR MULTI-AGENT PRODUCT SYSTEMS , 2002 .

[7]  Michael G. Safonov,et al.  The unfalsified control concept and learning , 1997 .

[8]  László Monostori,et al.  AI and machine learning techniques for managing complexity, changes and uncertainties in manufacturing , 2003 .

[9]  Agostino Poggi,et al.  Multiagent Systems , 2006, Intelligenza Artificiale.

[10]  Robert L. Kosut,et al.  Adaptive robust control: online learning , 1991, [1991] Proceedings of the 30th IEEE Conference on Decision and Control.

[11]  Håkan Hjalmarsson,et al.  For model-based control design, closed-loop identification gives better performance , 1996, Autom..

[12]  Nicholas R. Jennings,et al.  Agent-based control systems: Why are they suited to engineering complex systems? , 2003 .

[13]  Nicholas R. Jennings,et al.  Agent-based control systems , 2003 .

[14]  Pedro Albertos,et al.  Iterative Identification and Control , 2002, Springer London.

[15]  Hao Xia,et al.  Making use of instability and input saturation in a convergent unfalsification based iterative scheme , 1999 .

[16]  Sandor M. Veres.  Exploratory modelling for controller optimization , 2003 .

[17]  Raymond A. de Callafon,et al.  Identification of Normalised Coprime Plant Factors from Closed-loop Experimental Data , 1995, Eur. J. Control.

[18]  Sandor M. Veres,et al.  Synergy and duality of identification and control , 1999 .

[19]  Sandor M. Veres.  Convergence of control performance by unfalsification of models - levels of confidence , 2001 .

[20]  O.H. Bosgra,et al.  Identification of normalized coprime plant factors for iterative model and controller enhancement , 1993, Proceedings of 32nd IEEE Conference on Decision and Control.

[21]  Michel Gevers,et al.  Towards a Joint Design of Identification and Control , 1993 .