Self-organizing Approximation-based Control

University Of California-Riverside, Riverside CA

Investigators

Abstract

Proposal Number: 0701791 Proposal Title: Self-organizing Approximation-based Control PI Name: Farrell, Jay A. PI Institution: University of California-Riverside The objective of this research is to investigate and develop self-organizing approximation-based control methods applicable to physical systems with nonlinear dynamics and model uncertainty. Example applications include: autonomous aircraft subject to aging and damage; low-cost, agile, autonomous underwater vehicles; respirator air volume control; and, internal combustion engine control. As the performance specified for such systems increases, the nonlinear effects and model uncertainty become increasingly important. The approach of this research is to approximate the uncertain nonlinear dynamics during operation using self-organized approaches wherein the structure of the function approximator is defined by the control system during operation, with new approximator resources allocated as necessary to achieve a performance specification. The concept is analogous to the growth of the nervous system within a biological entity as it learns new motor-control tasks. Intellectual Merit: This research is applicable to high-performance control of nonlinear systems with model uncertainty. Self-organized approximation-based control allows more efficient implementations in applications with low-dimensional input spaces and can enable new applications where the input space dimension is currently too large for practical implementation using fixed-structure approximators. Broader Impacts: The research team has collaborative relationships with military and industry that includes direct participation in experiments. This collaboration enables communication of research results to those colleagues and of open questions back to the university research team. Research results are also published in appropriate venues for communication to the general research community. The project directly supports the education of graduate students. Female and under-represented categories of students are encouraged to participate.