Real-Time Model Predictive Control

Lehigh University Office of Technology Transfer
posted on 01/14/2008

• request more info 
•   
•  LICENSE ONLINE

Description:

This technology combines Model Predictive Control (MPC) with System-on-a-chip (SoC). This combination makes MPC applicable to portable devices.  The approach is the first reported implementation of real-time MPC on-a-chip. Current software implementations of MPC are not suitable for portable applications due to their relatively slow performance and high power consumption. Lehigh has developed a co-processor which acts as an accelerator to a general purpose microprocessor. This technology is suitable for applications with fast sampling time and small size, including diabetes testing, avionics, and automotive controllers.

A utility patent application US 11/876,440 has been filed on this technology.

The Market:

The global processing control market was estimated to be $601.5 million in 2005. Process control takes up around 45% of this overall market, or $270.6 million. [1] Within the on-chip and portable device market, manufacturers are looking to integrate established and existing controller technologies, such as MPC, and methodologies to take advantage of new market opportunities and increase functionality and performance. [2] At the same time, MPC applications are expanding as their functionality in more complex processes are being realized. [3]

 

The Opportunity:

Lehigh is interested in licensing this technology. 

 

About Lehigh University:

Lehigh University is a leading research university with an enrollment of approximately 8,000 undergraduate and graduate students. 

 

For More Information on this Technology Contact:

Yatin S. Karpe, Ph.D.
Associate Director

Office of Technology Transfer
Lehigh University
354 Whitaker Laboratory
5 East Packer Ave.
Bethlehem, PA 18015

Phone: (610) 758-5841
Fax: (610) 758-5888 fax
Email: yatin.karpe@lehigh.edu
Web: http://www.lehigh.edu/ott

Claire Casola

Licensing Coordinator

Foresight Science & Technology

Phone & Email: 401-273-4844 ext. 17; claire.casola@foresightst.com

http://www.foresightst.com/index.html

---

]]>

The proposed approach provides the first reported implementation of real-time model predictive (MPC) control on-a-chip. Current software implementations of MPC are not suitable for demanding applications due to their relatively slow performance and high power consumption. We proposed a custom made co-processor that acts as an accelerator to a general purpose microprocessor, both operating at reduced precision. The co-processor utilizes the logarithmic number system to perform the arithmetic operations which is proven to consume less power and performs faster than alternatives at reduced precision (such as the standard fixed-point and floatingpoint arithmetic).