DEPSCOR: Research on ARL's Intelligent Control Architecture

Grants and Contracts Details


~e propose extensions of th~ ongoing collaborative work on development of a Prototype IntellIgent Control (PIC) archItecture. Research on the PIC architecture has been condu~ ted ?y the researchers at the Applied Research Laboratory (ARL) of Pennsylvania S.,nte UmversIty, under the leadership of Dr. James A. Stover, and has been supported by the Office of Naval Research (ONR). PI's collaboration with ARL researchers started when he first visited ARL on a sabbatical leave during 1997-98, and since then he has been invited to ARL once every six month. The PIC architecture has been developed to provide a canonical architecture that is modular and hierarchical, and bears similalrity to biological systems, in order to simplify the design of "intelligent" controllers for autonomous systems operating in uncertain environments. Conventional controller designs are best suited for systems that have good models available, that operate in relatively known environments, and that have traditional control objectives such as stabilization, tracking, disturbance rejection, model matching, etc~ The so called intelligent controllers are used when either a good system' model is not available, or the control objectives are beyond the scope of conventional controllers. Past research at ARL has resulted in a control architecture, called the PIC architecture, that facilitates simplified design of intelligent controllers for complex systems. The simplicity of the designed controller results from the inherent modular and hierarchical structure required by th~ architecture, and its choice of internal biologically motivated functional blocks. Intelligent controllers designed within this architecture have already been successfully implemented in naval and medical applications. Further research is needed to enhance the capability of the intelligent controllers designed within this architecture. Some enhancements we propose to undertake include (i) Inclusion of randomness uncertainty (captured by probability measures) besides the already present vagueness uncertainty (captured by fuzzy logic measures); (ii) Inclusion of learning and adaptation capability; (iii) Modification of fuzzy logic connective so that they are algebraically better behaved and study the implications; (iv) Validation of these enhancements through simulation using data provided by the ARL. Before presenting the proposed research, we first describe the PIC, a behavior-based intelligent control architecture. This architecture yields an intelligent controller which is a cascade of a perceptor, and a responder. The perceptor builds internal representations as object class instances of the external world as detected by the sensors. The existence of class instances with certain properties enables the responder to react to them by appropriate behaviors. The behavioral approach additionally yields a hierarchical two lay~red responder, thus facilitating a better management of complexity. The responder uses the perceptor inputs to first compute the higher level activities, called behaviors, and next the corresponding lower level activities, called actions. Our prior work summarizes this in [KS98a, KSK98, SHG96, KS99, SK99, KS98b]. Application of the intelligent control architecture to underwater vehicles is reported in [ZAHV99], to anesthesiology in [SM99], and to coordination of mutiple underwater vehicles in [KS98b].
Effective start/end date4/1/0112/31/05


  • Office of Naval Research: $232,979.00


Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.