You are here

Embedding of Advanced Search Technique for Detect, Locate, and Track for Pedestrian-based Search

Award Information
Agency: Department of Homeland Security
Branch: N/A
Contract: HSHQDC-12-C-00108
Agency Tracking Number: DNDOSBIR12-02-FP-001-IOS
Amount: $149,942.65
Phase: Phase I
Program: SBIR
Solicitation Topic Code: 12.1-002
Solicitation Number: N/A
Solicitation Year: 2012
Award Year: 2012
Award Start Date (Proposal Award Date): 2012-09-18
Award End Date (Contract End Date): 2013-03-17
Small Business Information
Torrance, CA 90505-5217
United States
DUNS: 033449757
HUBZone Owned: No
Woman Owned: No
Socially and Economically Disadvantaged: No
Principal Investigator
 Andy Gooden
Business Contact
 Reuben Sandler
Research Institution

Techniques employed by expert operators to enhance the ability to detect, locate and
track radiation sources are straightforward in concept but require substantial skill in
execution and in interpretation of the detector readings obtained, combined with the
operator's visual observations. By combining detectors with fast acquisition times, below
500 milliseconds, with the video and computational capabilities contained in low cost
COTS systems such as smart phones, tablets, or other embedded processors, algorithms
can be developed that enhance detector components with improved capabilities and make
these expert search techniques even more effective. Additionally, although not a primary
goal, these expert techniques will be accessible to a much larger population of operators
and could reduce training costs.
Enabling an automated system to duplicate the performance of an expert operator is a
sensor data fusion problem. The automated system must acquire data from a set of
sensors and combine the data in a manner that at least matches the effectiveness of the
expert operator. For the current problem this means fusing location, bearing, video
(visual), motion and radiation data. Of these, video sensors present a unique challenge
because of the very large and noisy data they present, requiring the use of sophisticated
feature extraction techniques. Intelligent Optical Systems' approach will leverage the
increased computing power now available on COTS platforms as well as existing
software resources such as OpenCV or Kinect for Windows for video processing to
address this challenge. The data provided by other sensors, including accelerometers,
magnetometers and GPS can be converted to a useful form by relatively straightforward
software operation. After extraction of video features the input data, including video,
bearing, motion, location and radiation will be combined using a particle filter
implementation of a dynamic Bayesian network. This has been shown to be an effective
technique for fusing various types of input data, specifically including radiation and
location data. We will innovate to extend the algorithms to include extracted video,
bearing and motion data.
Through this program we will produce a set of products that provide a significant
enhancement to the performance of COTS detectors at low cost. Selected algorithms,
applications, visualization and computational capabilities of the system will be
implemented in a breadboard prototype at Technology Readiness Level TRL 4 to TRL 5,
and investigated in Phase I. Based on these results in Phase II the research and
development effort will be both broadened, to encompass additional COTS components,
and deepened to more fully investigate the optimum methods of enhancing performance,
reaching TRL 8 at the conclusion of Phase II. In total the products developed under this
SBIR will provide an extremely cost effective method to enhance the capabilities of
existing radiation detectors.
Intelligent Optical Systems’ team for this investigation has unique recent experience with
this subject matter and a broad background in algorithms and techniques for detection,
location and tracking of radiation sources as well as networking protocols, handheld
electronics, and embedded software.

* Information listed above is at the time of submission. *

US Flag An Official Website of the United States Government