Advanced MEC Discrimination Comparative Study on Standardized Test-Site Data Using Linear Genetic Programming (LGP) Discrimination
Dr. Dean Keiswetter | Acorn SI
In 2003, the Defense Science Board observed: “The … problem is that instruments that can detect the buried unexploded ordnance (UXO) also detect numerous scrap metal objects and other artifacts, which leads to an enormous amount of expensive digging. Typically 100 holes may be dug before a real UXO is unearthed! The Task Force assessment is that much of this wasteful digging can be eliminated by the use of more advanced technology instruments that exploit modern digital processing and advanced multi-mode sensors to achieve an improved level of discrimination of scrap from UXO.” Significant progress has been made in discrimination technology since that time. To date, these technologies have primarily been tested at constructed test sites, with only limited application at live sites. The routine implementation of discrimination technologies will require demonstrations at real UXO sites under real world conditions.
Objectives of the Demonstration
The objective of this project was to advance and improve munitions and explosives of concern (MEC) discrimination performance by validating a decision process that combines statistical analyses of digital geophysical mapping (DGM) products and linear genetic programming (LGP) methods to enable classification and provide iterative quantitative residual risk assessments that may be used during the excavation phase to determine a stop-digging cutoff.
Advanced classification was demonstrated at two live sites— the former Camp Sibert, Alabama and the former Camp San Luis Obispo (SLO), California. For data acquired at Camp Sibert, no iterations were required because the original classification was nearly perfect. At Camp SLO, the sampling of additional ground truth for a second iteration of discrimination and risk analysis improved the performance of the technology over the first iteration by almost any metric. In other words, intelligently selecting which targets to “dig” and then rebuilding discrimination models using those new targets as training targets improved UXO discrimination results and the accuracy of the project team’s residual risk assessment.
This project demonstrated an improvement in current best practices for discriminating MEC from clutter and the ability to produce principled site-closure decisions in a cost-effective manner. The approach relies on accurate three-dimensional spatial measurements as well as on stable geophysical measurements. Measurement of the altitude of the geophysical sensor is important to inverting for meaningful model parameters. If the data going into the inversion routines are noisy or contain systemic problems, final discrimination decisions will not be acceptable.