Embedded Biomorphic Chips for Focal Plane Imaging Processing and Biped Locomotion Control

Ralph Etienne-Cummings
John Hopkins University

Time and Place

1305 Newell-Simon Hall
Refreshments 3:15 pm
Talk 3:30 pm

Abstract

Embedded image processing on the focal plane promises high image signal quality and fast, compact and low-power information extraction for machine vision applications. Over the past decade, the promise of improved signal quality has been achieved, and we are now able to buy cheap ~60dB CMOS high-resolution imagers. The goal of fast, compact and low-power information extraction is, however, still elusive. Problems with inefficient chip area usage, incompatible algorithms and high power consumption, typically high speed digital, have limited the success on this front. We have developed a technique that attempts to provide all these promises using digitally programmable analog computation. Our approach remains compatible with standard computer vision interfaces by preserving the image resolution and scanning format, while in parallel extracting various spatiotemporal features from images. Furthermore, our systems performs greater than 10 GOPS/mW, albeit at ~6 bits of resolution, at many thousands of frames per second. We use a block parallel, mixed signal MIMD architecture with low silicon footprint for this purpose. I will present the evolution of our General Image Processing (GIP) chip and show its potential power as an Adaptive Computational Sensor (ACS) for synthetic imaging. I will describe the processing architecture and demonstrate its application to spatiotemporal filtering and line segment orientation detection. I will also describe a biomorphic chip for the adaptive control of biped locomotion. I will show some sensory adaptation examples.

Speaker Biography

Ralph Etienne-Cummings received his B. Sc. in physics, 1988, from Lincoln University, Pennsylvania. Complete his M.S.E.E. and Ph.D. in electrical engineering at the University of Pennsylvania in 1991 and 1994, respectively. Currently, Dr. Etienne-Cummings is an assistant professor of electrical and computer engineering at the Johns Hopkins University, Baltimore, MD. His research interest includes mixed signal VLSI systems, computational sensors, computer vision, neuromorphic engineering, smart structures, mobile robotics and robotics-assisted surgery. He is a recipient of the NSFUs Career Development and ONR's YIP Awards.

Speaker Appointments

For appointments, please contact the host, Vladimir Brajovic (brajovic@cs.cmu.edu).