Details

Project TitleSequential Action Control for Efficient Predictive Optimal Control
Track Code2014-164
Short Description

Sequential Action Control (SAC) as a computationally efficient optimal control algorithm for broad classes of linear and nonlinear systems

#software #optimization

Abstract

Efficient optimal control of nonlinear systems, such as complex robotics systems, is critical to the proper functioning of these systems. Traditional methods iteratively optimize the control sequences subject to constraints to minimize trajectory objectives. Such iterative process requires expensive computational resources and is inefficient for real-time control of complicated systems. To overcome these issues, Northwestern researchers have recently developed an innovative method called Sequential Action Control (SAC) for the efficient predictive optimal control of both linear and nonlinear systems. SAC avoids the expensive iterative process by computing the infinitesimal optimum at each instant to optimally improve the system trajectory objective. This fundamental advantage enables SAC to provide real-time, on-line, closed-loop optimal control for nonlinear system many orders of magnitude faster than other conventional methods. Allowing the benefits of predictive optimal control to be applied on-line and in closed-loop for systems where such methods would normally prove infeasible, SAC help to drive new classes of increasingly complex processes and robots to the limit of their capabilities.

 
Tagssoftware, software: optimization
 
Posted DateFeb 20, 2015 3:04 PM

Inventor(s)

Todd D. Murphey

Alexander R. Ansari

Applications

·         Real-time control for auto-pilot and stabilization systems,

·         Predictive feedback control for complex and flexible robotic system

·         Real-time adaptive control and automation

·         On-line high frequency control of nonlinear system

Advantages

·         Low computational resource requirement

·         Super-fast On-line optimal control for nonlinear system

·         Robust against local optimum

Publications

IP Status

US provisional patent has been filed.

Marketing Contact

Arjan Quist, PhD

Invention Associate

(p) 847-467-0305

(e) arjan.quist@northwestern.edu

Files

File Name Description
2014-164 Marketing None Download