Distributed Receding Horizon Control
|Prev: Cooperative and Coordinated Control Scheme for Multi-Agent Systems||Course Home||Next: CDS 270-2, Spring 2006|
This lecture will illustrate how the tools described in the Wednesday 04/12/06 lecture can be used to study stability of decentralized receding horizon control (DRHC). We will discuss why the problem is more difficult and what are the main issues that drive the problem. Stability of DRHC will be studied in a discrete-time framework for the class of dynamically decoupled systems. Alternative approaches and various practical methods to help constraint fulfillment will be mentioned as well.
Stability analysis of decentralized RHC for decoupled systems, T. Keviczky, F. Borrelli and G. J. Balas. 44th IEEE Conf. on Decision and Control, and European Control Conference, December 2005, Seville, Spain. This paper contains a description of the discrete-time distributed RHC framework discussed in class and gives sufficient conditions for stability.
Distributed receding horizon control for multi-vehicle formation stabilization, W. B. Dunbar and R. M. Murray. Automatica, 2006, Vol. 42, No. 4, pp. 549-558. This paper presents a continuous-time distributed RHC framework and uses exchange of open-loop optimal trajectories to achieve stability.
Distributed Model Predictive Control, E. Camponogara, D. Jia, B. H. Krogh and S. Talukdar. IEEE Control Systems Magazine, February 2002. This is one of the earlier works considering distributed RHC for dynamically coupled LTI subsystem dynamics with quadratic separable cost functions.
Decentralized Receding Horizon Control of Large Scale Dynamically Decoupled Systems, T. Keviczky. PhD Thesis, September 2005.
Distributed Receding Horizon Control of Multiagent Systems, W. B. Dunbar. PhD Thesis, April 2004.
A Decentralized Algorithm for Robust Constrained Model Predictive Control, A. Richards and J. How. American Control Conference, June 2004, Boston, Massachusetts. This paper uses robust constraint fulfillment for a hierarchical interconnection graph where the exchange of optimal control policies occur sequentially. Feasibility of coupling constraints (e.g. collision avoidance) is addressed in this setting for linear, dynamically decoupled subsystems.