The overall goal of this project was to examine the technologies and processes involved in the remediation of unexploded ordnance (UXO), and then to provide recommendations on how each could be enhanced in the future. There were two primary objectives of this research: (1) to develop a better understanding of the environmental factors and phenomenologies that dictate success and failure of subsurface anomaly detection through a careful study of controlled field tests involving several sensor technologies; and (2) to identify the most promising data processing, analysis, and data integration (fusion) approaches to improve UXO detection, discrimination, and identification/classification through the fusion of multiple complementary data sets.
The phenomenology studies were the first element of this project. There were two aspects to these studies: (1) assessment of prior efforts, including the Defense Advanced Research Projects Agency (DARPA) Backgrounds Program UXO studies and the Jefferson Proving Ground (JPG), Indiana UXO demonstrations; and (2) the development of prototype forward models for magnetic and electromagnetic induction (EMI) signatures of UXO-like targets and comparison to measured signatures. The assessment of prior studies was accomplished by forming an honest-broker technical review committee made up of government scientists who would solicit genuine technical evaluations of their system's performance from several of the participants of the technology demonstrations. The review committee critiqued each of these self-examinations and prepared a report that would address many of the phenomenological issues that affect sensor performance in various environments and provided general guidance for system improvements that support an effort to develop a multi-sensor detection platform.
The second element of this project involved an examination of data integration (fusion) algorithm(s) and approaches, focused on the combination of data from several complementary sensors to optimize detection system performance. This effort began by examining more closely any attempts at data fusion performed by JPG demonstration participants who utilized multiple sensors. A prototype data fusion algorithm was then developed and tested on JPG data taken from both multiple sensor platforms and from multiple sensors used by different participants that could be spatially co-registered. The final product of this effort is a model that can be adapted to the fusion of data from the types of sensor systems that would serve as the prototype UXO detection system.
It was found that UXO detection, except for problematic cases, is presently possible with both single and multiple complementary geophysical datasets. The report findings concentrate on defining the current status of detection capability and then address the discrimination and identification problems. The two problematic cases for which detection capability does not currently exist: (1) small, widely spaced, single-ordnance items, such as 20-mm projectiles; and (2) ordnance items buried too deeply, relative to their size, for detection with present technology.
On the subject of discrimination, the research concluded that there is not sufficient information content in currently deployed single-sensor survey data to allow UXO discrimination, except in special cases. General discrimination capability will require multisensory, complementary datasets and integrated or joint interpretation. Additionally, it has been found that UXO identification or classification is not feasible with currently deployed technology. Limited, reported successes in UXO identification using empirical correlation of physical properties such as interpreted magnetic moment from test site surveys to measured magnetic moments for actual ordnance items cannot be extended to general surveys of real world UXO contaminated sites. A key component of the ultimate solution of the UXO discrimination and identification requirement is the development of high fidelity, physics-based models for forward and inverse modeling of geophysical signatures. Examples of what such models will “look like” are given in the final report, based on the development of “first generation” forward models of magnetic and EMI signatures of UXO-like targets and comparison to measured signatures.
The successful conduct of the elements of this research project have placed the U.S. Army Corps of Engineers in a position to assemble a prototype multi-sensor UXO detection platform along with the software necessary to accomplish fusion of the enormous amount of data such a platform will develop. Follow-on commercialization of this technology will provide a significant financial savings over current man-in-the-loop UXO detection and clearance operations that focus on areas that are limited in physical size.